Home Docs Claude Developer Platform

One capture · api

One read of Claude Developer Platformapi-20260821T200735Z

123 pages moved out of 584 read.

Pages moved 123 significant first

Pages read 584 in this capture

Captured 21 Aug 2026 20:07 UTC

Corpus hash 10824cc06631 index-hash

What this read moved

1-25 of 123, page 1 of 5

This capture is too large to show at once. Changes 1-25 of 123 are below, significant first; the rest are on the following screens.

about-claude/models/migration-guide Changed · +20 / -21 lines

from line 20

2020

2121## Migrating to Claude Mythos 5 and Claude Fable 5

2222

23[Claude Fable 5](https://platform.claude.com/docs/en/about-claude/models/introducing-claude-fable-5-and-claude-mythos-5) is Anthropic's most capable widely released model, generally available on the Claude API, [Amazon Bedrock](https://platform.claude.com/docs/en/build-with-claude/claude-in-amazon-bedrock), [Claude Platform on AWS](https://platform.claude.com/docs/en/build-with-claude/claude-platform-on-aws), [Google Cloud](https://platform.claude.com/docs/en/build-with-claude/claude-on-vertex-ai), and [Microsoft Foundry](https://platform.claude.com/docs/en/build-with-claude/claude-in-microsoft-foundry). [Claude Mythos 5](https://anthropic.com/glasswing) shares the same capabilities and is offered in limited availability to approved customers in Project Glasswing.

23[Claude Fable 5](https://platform.claude.com/docs/en/about-claude/models/introducing-claude-fable-5-and-claude-mythos-5) is Anthropic's most capable widely released model, available on the Claude API, [Amazon Bedrock](https://platform.claude.com/docs/en/build-with-claude/claude-in-amazon-bedrock), [Claude Platform on AWS](https://platform.claude.com/docs/en/build-with-claude/claude-platform-on-aws), [Google Cloud](https://platform.claude.com/docs/en/build-with-claude/claude-on-vertex-ai), and [Microsoft Foundry](https://platform.claude.com/docs/en/build-with-claude/claude-in-microsoft-foundry). [Claude Mythos 5](https://anthropic.com/glasswing) shares the same capabilities and is offered only to approved customers in Project Glasswing.

2424

2525The baseline settings shared by `claude-fable-5` and `claude-mythos-5`:

2626

from line 32

3232

3333Where the two models diverge:

3434

35* **Availability:** Claude Fable 5 is generally available. Claude Mythos 5 is available only to approved customers in [Project Glasswing](https://anthropic.com/glasswing).

35* **Availability:** Claude Fable 5 does not require access approval. Claude Mythos 5 is available only to approved customers in [Project Glasswing](https://anthropic.com/glasswing).

3636* **Safety classifiers:** Claude Fable 5 runs safety classifiers that can decline requests with `stop_reason: "refusal"`. Claude Mythos 5 does not include these classifiers. See [Refusals and fallback](https://platform.claude.com/docs/en/build-with-claude/refusals-and-fallback).

3737* **Priority Tier:** [Priority Tier](https://platform.claude.com/docs/en/api/service-tiers#supported-models) is supported on Claude Fable 5 but not on Claude Mythos 5.

3838

3939### Migrating to Claude Mythos 5 and Claude Fable 5 from Claude Mythos Preview

4040

41[Claude Mythos 5](https://anthropic.com/glasswing) is the access-gated successor to [Claude Mythos Preview](https://anthropic.com/glasswing), the invitation-only research preview. [Claude Fable 5](https://platform.claude.com/docs/en/about-claude/models/introducing-claude-fable-5-and-claude-mythos-5) is the generally available model with the same capabilities, and the changes in this section apply equally to both targets.

41[Claude Mythos 5](https://anthropic.com/glasswing) is the access-gated successor to [Claude Mythos Preview](https://anthropic.com/glasswing), the invitation-only research preview. [Claude Fable 5](https://platform.claude.com/docs/en/about-claude/models/introducing-claude-fable-5-and-claude-mythos-5) offers the same capabilities and does not require access approval. The changes in this section apply equally to both targets.

4242

4343Migration is mostly drop-in. Claude Mythos 5 and Claude Fable 5 use the same [Messages API](https://platform.claude.com/docs/en/build-with-claude/working-with-messages) and the same [tool use](https://platform.claude.com/docs/en/agents-and-tools/tool-use/overview) patterns as Claude Mythos Preview, and token counts are roughly unchanged because all three models use the same tokenizer. The key changes to check are the features that are no longer available (listed in the next section) and thinking output. If you migrate to Claude Fable 5, also plan for safety classifier refusals, which Claude Mythos Preview and Claude Mythos 5 do not have; see [Refusals and fallback](https://platform.claude.com/docs/en/build-with-claude/refusals-and-fallback).

4444

from line 50

5050model = "claude-mythos-preview" # Before

5151model = "claude-mythos-5" # After

5252

53# Or, for the generally available model with the same capabilities:

53# Or, for the model with the same capabilities and no access approval requirement:

5454model = "claude-fable-5" # After

5555```

5656

from line 319

319319

320320#### Migration checklist

321321

322* Update the model name from `claude-mythos-preview` to `claude-mythos-5`, or to `claude-fable-5` for the generally available model.

322* Update the model name from `claude-mythos-preview` to `claude-mythos-5`, or to `claude-fable-5`, which offers the same capabilities and does not require access approval.

323323* Remove manual extended thinking configuration (`thinking: {type: "enabled", budget_tokens: N}`). Adaptive thinking is always on, and no `thinking` field is required.

324324* Remove any `thinking: {type: "disabled"}` configuration. Disabling thinking returns an error on `claude-mythos-5` and `claude-fable-5`.

325325* Remove `budget_tokens`. It has no direct replacement: thinking is adaptive, and the `effort` parameter is a separate output-level control, not a thinking budget.

from line 794

794794

795795### Migrating to Claude Opus 5 from Claude Opus 4.7

796796

797Claude Opus 5 should have strong out-of-the-box performance on existing Claude Opus 4.7 prompts and evals, at the same pricing of $5 per million input tokens and $25 per million output tokens. It supports the same set of features as Claude Opus 4.7, including the [1M token context window](https://platform.claude.com/docs/en/build-with-claude/context-windows), [128k max output tokens](https://platform.claude.com/docs/en/about-claude/models/overview), [adaptive thinking](https://platform.claude.com/docs/en/build-with-claude/thinking), [prompt caching](https://platform.claude.com/docs/en/build-with-claude/prompt-caching), [batch processing](https://platform.claude.com/docs/en/build-with-claude/batch-processing), the [Files API](https://platform.claude.com/docs/en/build-with-claude/files), [PDF support](https://platform.claude.com/docs/en/build-with-claude/pdf-support), [vision](https://platform.claude.com/docs/en/build-with-claude/vision), and server-side and client-side [tools](https://platform.claude.com/docs/en/agents-and-tools/tool-use/overview), with two exceptions: [web fetch](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-fetch-tool) is not available on Claude Opus 5, and [Priority Tier](https://platform.claude.com/docs/en/api/service-tiers#supported-models) is not supported on Claude Opus 5. It also adds [mid-conversation system messages](https://platform.claude.com/docs/en/build-with-claude/mid-conversation-system-messages) and publicly documents [refusal stop details](https://platform.claude.com/docs/en/build-with-claude/refusals-and-fallback#refusal-response). On the Claude API, Claude Opus 5 also supports [computer use](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool) as the generally available `computer_toolset_20260801` toolset and the [browser use tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool) for tasks inside webpages, neither of which Claude Opus 4.7 supports; existing integrations on the earlier `computer_20251124` version continue to work unchanged on both models. To upgrade an existing integration, see [Migrate from `computer_20251124`](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool#migrate-from-computer-20251124).

797Claude Opus 5 should have strong out-of-the-box performance on existing Claude Opus 4.7 prompts and evals, at the same pricing of $5 per million input tokens and $25 per million output tokens. It supports the same set of features as Claude Opus 4.7, including the [1M token context window](https://platform.claude.com/docs/en/build-with-claude/context-windows), [128k max output tokens](https://platform.claude.com/docs/en/about-claude/models/overview), [adaptive thinking](https://platform.claude.com/docs/en/build-with-claude/thinking), [prompt caching](https://platform.claude.com/docs/en/build-with-claude/prompt-caching), [batch processing](https://platform.claude.com/docs/en/build-with-claude/batch-processing), the [Files API](https://platform.claude.com/docs/en/build-with-claude/files), [PDF support](https://platform.claude.com/docs/en/build-with-claude/pdf-support), [vision](https://platform.claude.com/docs/en/build-with-claude/vision), and server-side and client-side [tools](https://platform.claude.com/docs/en/agents-and-tools/tool-use/overview), with two exceptions: [web fetch](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-fetch-tool) is not available on Claude Opus 5, and [Priority Tier](https://platform.claude.com/docs/en/api/service-tiers#supported-models) is not supported on Claude Opus 5. It also adds [mid-conversation system messages](https://platform.claude.com/docs/en/build-with-claude/mid-conversation-system-messages) and publicly documents [refusal stop details](https://platform.claude.com/docs/en/build-with-claude/refusals-and-fallback#refusal-response). On the Claude API, Claude Opus 5 also supports [computer use](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool) as the stable `computer_toolset_20260801` toolset and the [browser use tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool) for tasks inside webpages, neither of which Claude Opus 4.7 supports; existing integrations on the earlier `computer_20251124` version continue to work unchanged on both models. To upgrade an existing integration, see [Migrate from `computer_20251124`](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool#migrate-from-computer-20251124).

798798

799799<Note>

800800 If your code is on Claude Opus 4.6 or earlier, use [Migrating to Claude Opus 5 from Claude Opus 4.6 and earlier Opus models](https://platform.claude.com/docs/en/about-claude/models/migration-guide#migrating-from-claude-opus-46) instead. That section includes breaking changes (sampling parameters rejected, manual extended thinking rejected, new tokenizer) that the upgrade from Claude Opus 4.7 alone does not cover.

from line 855

855855

856856The following items are not breaking changes; they describe behavior differences worth checking after you swap the model ID.

857857

8581. **Sampling parameters (unchanged):** Setting `temperature`, `top_p`, or `top_k` to a non-default value returns a 400 error on Claude Opus 5, the same as on Claude Opus 4.7. The SDK request types still define these fields for compatibility with earlier models, so code that sets them type-checks, but the API rejects the request server-side. If you removed these parameters when migrating to Opus 4.7, no further changes are needed.

8581. **Sampling parameters (unchanged):** Setting `temperature`, `top_p`, or `top_k` to a non-default value returns a 400 error on Claude Opus 5, the same as on Claude Opus 4.7. Most SDKs still define these fields for compatibility with earlier models, so code that sets them type-checks even though the API rejects the request. The Python SDK (v1.0 and later) does not define them, and passing them raises a `TypeError`. If you removed these parameters when migrating to Opus 4.7, no further changes are needed.

859859

8608602. **Effort default is `high`:** The [effort parameter](https://platform.claude.com/docs/en/build-with-claude/effort) default on Claude Opus 5 is `high` on the Claude API and Claude Code. If you already set effort explicitly, your setting is unchanged.

861861

from line 915

915915* [Vision](https://platform.claude.com/docs/en/build-with-claude/vision)

916916* Server-side and client-side [tools](https://platform.claude.com/docs/en/agents-and-tools/tool-use/overview) ([bash](https://platform.claude.com/docs/en/agents-and-tools/tool-use/bash-tool), [code execution](https://platform.claude.com/docs/en/agents-and-tools/tool-use/code-execution-tool), [computer use](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool), [text editor](https://platform.claude.com/docs/en/agents-and-tools/tool-use/text-editor-tool), [web search](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-search-tool), [MCP connector](https://platform.claude.com/docs/en/agents-and-tools/mcp-connector), [memory](https://platform.claude.com/docs/en/agents-and-tools/tool-use/memory-tool))

917917

918Two exceptions: [web fetch](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-fetch-tool) is not available on Claude Opus 5, and [Priority Tier](https://platform.claude.com/docs/en/api/service-tiers#supported-models) is not supported on Claude Opus 5. On the Claude API, Claude Opus 5 also supports [computer use](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool) as the generally available `computer_toolset_20260801` toolset and the [browser use tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool) for tasks inside webpages, neither of which Claude Opus 4.6 or earlier Opus models support; existing integrations on the earlier `computer_20251124` version continue to work unchanged on Claude Opus 5. To upgrade an existing integration, see [Migrate from `computer_20251124`](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool#migrate-from-computer-20251124).

918Two exceptions: [web fetch](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-fetch-tool) is not available on Claude Opus 5, and [Priority Tier](https://platform.claude.com/docs/en/api/service-tiers#supported-models) is not supported on Claude Opus 5. On the Claude API, Claude Opus 5 also supports [computer use](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool) as the stable `computer_toolset_20260801` toolset and the [browser use tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool) for tasks inside webpages, neither of which Claude Opus 4.6 or earlier Opus models support; existing integrations on the earlier `computer_20251124` version continue to work unchanged on Claude Opus 5. To upgrade an existing integration, see [Migrate from `computer_20251124`](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool#migrate-from-computer-20251124).

919919

920920#### Update your model name

921921

from line 1216

12161216

121712173. **Disabling thinking is capped at `high` effort:** You can turn thinking off with `thinking: {type: "disabled"}`, but only at an [effort](https://platform.claude.com/docs/en/build-with-claude/effort) level of `high` or below. A request that combines `thinking: {type: "disabled"}` with effort `xhigh` or `max` returns a 400 error on Claude Opus 5, enforced on each request. Audit requests that disable thinking before you migrate: re-enable thinking or lower the effort to `high` or below.

12181218

12194. **Sampling parameters removed:** Setting `temperature`, `top_p`, or `top_k` to any non-default value on Claude Opus 4.7 or later models, including Claude Opus 5, returns a 400 error. The safest migration path is to omit these parameters entirely from request payloads. Prompting is the recommended way to guide model behavior on Claude Opus 5. If you were using `temperature = 0` for determinism, note that it never guaranteed identical outputs on prior models.

12194. **Sampling parameters removed:** Setting `temperature`, `top_p`, or `top_k` to any non-default value on Claude Opus 4.7 or later models, including Claude Opus 5, returns a 400 error. The Python SDK (v1.0 and later) does not define them, and passing them raises a `TypeError`. The safest migration path is to omit these parameters entirely from request payloads. Prompting is the recommended way to guide model behavior on Claude Opus 5. If you were using `temperature = 0` for determinism, note that it never guaranteed identical outputs on prior models.

12201220

122112215. **Thinking content omitted by default:** Thinking blocks still appear in the response stream on Claude Opus 4.7 and later models, but their `thinking` field is empty unless you explicitly opt in. This is a silent change from Claude Opus 4.6, where the default was to return summarized thinking text. To restore summarized thinking content, set `thinking.display` to `"summarized"`:

12221222

from line 1624

16241624 ```

16251625 </CodeGroup>

16261626

1627 Note that the migration also moves from `client.beta.messages.create` to `client.messages.create`. Adaptive thinking and effort are GA features and do not require the beta SDK namespace or any beta headers.

1627 Note that the migration also moves from `client.beta.messages.create` to `client.messages.create`. Adaptive thinking and effort do not require the beta SDK namespace or any beta headers.

16281628

16292. **Remove effort beta header:** The effort parameter is now GA. Remove `betas=["effort-2025-11-24"]` from your requests.

16292. **Remove effort beta header:** The effort parameter does not require a beta header. Remove `betas=["effort-2025-11-24"]` from your requests.

16301630

16313. **Remove fine-grained tool streaming beta header:** Fine-grained tool streaming is now GA. Remove `betas=["fine-grained-tool-streaming-2025-05-14"]` from your requests.

16313. **Remove fine-grained tool streaming beta header:** Fine-grained tool streaming does not require a beta header. Remove `betas=["fine-grained-tool-streaming-2025-05-14"]` from your requests.

16321632

163316334. **Remove interleaved thinking beta header:** Adaptive thinking automatically enables interleaved thinking on Claude Opus 4.7, Opus 4.6, and Sonnet 4.6. Remove `betas=["interleaved-thinking-2025-05-14"]` from your requests. The header is still functional on Sonnet 4.6 with manual extended thinking, but manual mode is deprecated.

16341634

16355. **Migrate to output\_config.format:** If using structured outputs, update `output_format={...}` to `output_config={"format": {...}}`. The old parameter remains functional but is deprecated and will be removed in a future model release.

16355. **Migrate to output\_config.format:** If using structured outputs, update `output_format={...}` to `output_config={"format": {...}}`. The API still accepts the deprecated `output_format` parameter, but it will be removed in a future model release. The Python SDK (v1.0 and later) does not accept `output_format={...}` on `client.beta.messages.create()` or `count_tokens()`. The `output_format=Model` argument of the `parse()` and `stream()` helpers is unchanged.

16361636

16371637#### Migrating from Claude 4.1 or earlier

16381638

from line 1656

16561656 This is a breaking change when migrating from Claude 3.x models.

16571657 </Warning>

16581658

1659 Starting with Claude Opus 4.7, setting `temperature`, `top_p`, or `top_k` to any non-default value returns a 400 error. The safest migration path is to omit these parameters entirely from requests, and to use prompting to guide the model's behavior. If you were using `temperature = 0` for determinism, note that it never guaranteed identical outputs.

1659 Starting with Claude Opus 4.7, setting `temperature`, `top_p`, or `top_k` to any non-default value returns a 400 error. The Python SDK (v1.0 and later) does not define them, and passing them raises a `TypeError`. The safest migration path is to omit these parameters entirely from requests, and to use prompting to guide the model's behavior. If you were using `temperature = 0` for determinism, note that it never guaranteed identical outputs.

16601660

16611661 <CodeGroup exclude="shell">

16621662 ```python Python

from line 1997

19971997* **BREAKING:** Remove assistant message prefills (returns 400 error); use structured outputs or `output_config.format` instead

19981998* **BREAKING on Opus 4.7:** Replace `thinking: {type: "enabled", budget_tokens: N}` with `thinking: {type: "adaptive"}` plus the [effort parameter](https://platform.claude.com/docs/en/build-with-claude/effort) (returns 400 on Opus 4.7)

19991999* Verify tool call JSON parsing uses a standard JSON parser

2000* Remove `effort-2025-11-24` beta header (effort is now GA)

2000* Remove `effort-2025-11-24` beta header (the effort parameter does not require it)

20012001* Remove `fine-grained-tool-streaming-2025-05-14` beta header

20022002* Remove `interleaved-thinking-2025-05-14` beta header (adaptive thinking enables interleaved thinking automatically)

20032003* Migrate `output_format` to `output_config.format` (if applicable)

from line 2027

20272027

202820282. **Disabling thinking is capped at `high` effort:** On Claude Sonnet 5, `thinking: {type: "disabled"}` is accepted at any effort level. On Claude Opus 5, it is accepted only at an [effort](https://platform.claude.com/docs/en/build-with-claude/effort) level of `high` or below; a request that combines `thinking: {type: "disabled"}` with effort `xhigh` or `max` returns a 400 error, enforced on each request. Audit requests that disable thinking before you migrate.

20292029

20303. **Mid-conversation system messages:** Claude Opus 5 accepts `role: "system"` messages immediately after a user turn in the `messages` array (subject to [placement rules](https://platform.claude.com/docs/en/build-with-claude/mid-conversation-system-messages#limitations)); Claude Sonnet 5 does not. If you maintain code paths that rebuild the full message history to update instructions, you can simplify them and preserve [prompt cache](https://platform.claude.com/docs/en/build-with-claude/prompt-caching) hits on earlier turns.

20303. **Mid-conversation system messages:** Claude Opus 5 accepts `role: "system"` messages immediately after a user turn in the `messages` array (subject to [placement rules](https://platform.claude.com/docs/en/build-with-claude/mid-conversation-system-messages#limitations)). This feature is not available on Claude Sonnet 5; use the top-level `system` field instead. If you maintain code paths that rebuild the full message history to update instructions, you can simplify them and preserve [prompt cache](https://platform.claude.com/docs/en/build-with-claude/prompt-caching) hits on earlier turns.

20312031

203220324. **Web fetch is not available:** The [web fetch](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-fetch-tool) tool is available on Claude Sonnet 5 but not on Claude Opus 5.

20332033

from line 2044

20442044

20452045Claude Sonnet 5 offers the best combination of speed and intelligence in the Claude model family. It builds on Claude Sonnet 4.6.

20462046

2047Claude Sonnet 5 is a drop-in upgrade for Claude Sonnet 4.6, priced at $2/$10 USD per million input/output tokens; see [Pricing](https://platform.claude.com/docs/en/about-claude/pricing) for details. There are two breaking API changes for code already running on Claude Sonnet 4.6: manual extended thinking (`thinking: {type: "enabled", budget_tokens: N}`) and sampling parameters (`temperature`, `top_p`, `top_k`) set to non-default values are no longer accepted and return a 400 error. Use [adaptive thinking](https://platform.claude.com/docs/en/build-with-claude/thinking) with the [effort parameter](https://platform.claude.com/docs/en/build-with-claude/effort) instead. Claude Sonnet 5 supports the same set of features as Claude Sonnet 4.6, including the [1M token context window](https://platform.claude.com/docs/en/build-with-claude/context-windows), [adaptive thinking](https://platform.claude.com/docs/en/build-with-claude/thinking), [prompt caching](https://platform.claude.com/docs/en/build-with-claude/prompt-caching), [batch processing](https://platform.claude.com/docs/en/build-with-claude/batch-processing), the [Files API](https://platform.claude.com/docs/en/build-with-claude/files), [PDF support](https://platform.claude.com/docs/en/build-with-claude/pdf-support), [vision](https://platform.claude.com/docs/en/build-with-claude/vision), and the full set of server-side and client-side [tools](https://platform.claude.com/docs/en/agents-and-tools/tool-use/overview). On the Claude API, Claude Sonnet 5 also supports [computer use](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool) as the generally available `computer_toolset_20260801` toolset and the [browser use tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool) for tasks inside webpages, neither of which Claude Sonnet 4.6 supports; existing integrations on the earlier `computer_20251124` version continue to work unchanged on both models. To upgrade an existing integration, see [Migrate from `computer_20251124`](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool#migrate-from-computer-20251124). [Priority Tier](https://platform.claude.com/docs/en/api/service-tiers#supported-models) is not available on Claude Sonnet 5. Claude Sonnet 5 also uses a new tokenizer.

2047Claude Sonnet 5 is a drop-in upgrade for Claude Sonnet 4.6, priced at $2/$10 USD per million input/output tokens; see [Pricing](https://platform.claude.com/docs/en/about-claude/pricing) for details. There are two breaking API changes for code already running on Claude Sonnet 4.6: manual extended thinking (`thinking: {type: "enabled", budget_tokens: N}`) and sampling parameters (`temperature`, `top_p`, `top_k`) set to non-default values are no longer accepted and return a 400 error. Use [adaptive thinking](https://platform.claude.com/docs/en/build-with-claude/thinking) with the [effort parameter](https://platform.claude.com/docs/en/build-with-claude/effort) instead. Claude Sonnet 5 supports the same set of features as Claude Sonnet 4.6, including the [1M token context window](https://platform.claude.com/docs/en/build-with-claude/context-windows), [adaptive thinking](https://platform.claude.com/docs/en/build-with-claude/thinking), [prompt caching](https://platform.claude.com/docs/en/build-with-claude/prompt-caching), [batch processing](https://platform.claude.com/docs/en/build-with-claude/batch-processing), the [Files API](https://platform.claude.com/docs/en/build-with-claude/files), [PDF support](https://platform.claude.com/docs/en/build-with-claude/pdf-support), [vision](https://platform.claude.com/docs/en/build-with-claude/vision), and the full set of server-side and client-side [tools](https://platform.claude.com/docs/en/agents-and-tools/tool-use/overview). On the Claude API, Claude Sonnet 5 also supports [computer use](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool) as the stable `computer_toolset_20260801` toolset and the [browser use tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool) for tasks inside webpages, neither of which Claude Sonnet 4.6 supports; existing integrations on the earlier `computer_20251124` version continue to work unchanged on both models. To upgrade an existing integration, see [Migrate from `computer_20251124`](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool#migrate-from-computer-20251124). [Priority Tier](https://platform.claude.com/docs/en/api/service-tiers#supported-models) is not available on Claude Sonnet 5. Claude Sonnet 5 also uses a new tokenizer.

20482048

20492049### Migrating to Claude Sonnet 5 from Claude Sonnet 4.6

20502050

from line 2104

21042104 ```

21052105

21062106 ```bash CLI

2107 ant messages create \

2108 --transform content --format yaml <<'YAML'

2107 ant messages create --transform content --format yaml <<'YAML'

21092108 model: claude-sonnet-5

21102109 max_tokens: 16000

21112110 thinking:

about-claude/models/optimizing-for-cost-and-intelligence Changed · +104 / -29 lines

from line 70

7070

7171Those breakpoint placements follow the standard pattern in [Explicit cache breakpoints](https://platform.claude.com/docs/en/build-with-claude/prompt-caching#explicit-cache-breakpoints).

7272

73Three settings can break your cache during a task. Changing [`effort`](https://platform.claude.com/docs/en/build-with-claude/effort) between requests invalidates the cached prefix, so change it only where you would re-cache anyway, such as at a [compaction](https://platform.claude.com/docs/en/build-with-claude/compaction) boundary. Changing a [task budget](https://platform.claude.com/docs/en/build-with-claude/task-budgets) partway through does the same, so set it once, on the first request. Every [context editing](https://platform.claude.com/docs/en/build-with-claude/context-editing#context-editing-and-prompt-caching) pass invalidates the prefix from the point it clears and the next request pays to re-cache everything after it, so clear in a few large batches rather than many small ones. Make all three changes at natural breaks, then confirm cache reads have not dropped; if they have, [cache diagnostics](https://platform.claude.com/docs/en/build-with-claude/cache-diagnostics) shows where the prefix diverged.

73Four things can break your cache during a task. Anything that changes per request, such as a timestamp or a queue position, placed ahead of the stable prefix turns every request into a full cache write: on the triage run in [Trim input and context tokens](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#trim-input-and-context-tokens), a 25-token status line at the front of the system prompt cost $4.24 per run instead of $0.59, more than running with caching off. Keep per-request text in the newest user turn. Changing [`effort`](https://platform.claude.com/docs/en/build-with-claude/effort) between requests invalidates the cached prefix, so change it only where you would re-cache anyway, such as at a [compaction](https://platform.claude.com/docs/en/build-with-claude/compaction) boundary. Changing a [task budget](https://platform.claude.com/docs/en/build-with-claude/task-budgets) partway through does the same, so set it once, on the first request. Every [context editing](https://platform.claude.com/docs/en/build-with-claude/context-editing#context-editing-and-prompt-caching) pass invalidates the prefix from the point it clears and the next request pays to re-cache everything after it, so clear in a few large batches rather than many small ones. Make the last three changes at natural breaks, then confirm cache reads have not dropped; if they have, [cache diagnostics](https://platform.claude.com/docs/en/build-with-claude/cache-diagnostics) shows where the prefix diverged.

7474

7575### Trim input and context tokens

7676

7777Most agent requests carry tokens that never influence the answer. Trimming them costs nothing in output quality, although not every lever here saved money when measured. Two places to look:

7878

79* **Input trimming.** [Dynamic filtering](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-fetch-tool#dynamic-filtering) in the web fetch tool keeps boilerplate out of fetched pages, [image resizing](https://platform.claude.com/docs/en/build-with-claude/vision#evaluate-image-size) right-sizes vision inputs, and [tool search with deferred loading](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-search-tool) loads tool definitions only when needed. [Programmatic tool calling](https://platform.claude.com/docs/en/agents-and-tools/tool-use/programmatic-tool-calling) lets Claude run several tool calls from code so only the filtered result enters the context; its documentation reports 24% fewer input tokens on agentic search benchmarks, with a higher score. [Manage tool context](https://platform.claude.com/docs/en/agents-and-tools/tool-use/manage-tool-context) compares tool search, programmatic tool calling, prompt caching, and context editing.

79* **Input trimming.** [Dynamic filtering](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-fetch-tool#dynamic-filtering) in the web fetch tool keeps boilerplate out of fetched pages, [image resizing](https://platform.claude.com/docs/en/build-with-claude/vision#evaluate-image-size) right-sizes vision inputs, and [tool search with deferred loading](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-search-tool) loads tool definitions only when needed (measured later in this section). [Programmatic tool calling](https://platform.claude.com/docs/en/agents-and-tools/tool-use/programmatic-tool-calling) lets Claude run several tool calls from code so only the filtered result enters the context; its documentation reports 24% fewer input tokens on agentic search benchmarks, with a higher score. [Manage tool context](https://platform.claude.com/docs/en/agents-and-tools/tool-use/manage-tool-context) compares tool search, programmatic tool calling, prompt caching, and context editing.

8080* **Context lifecycle.** [Context editing](https://platform.claude.com/docs/en/build-with-claude/context-editing) clears stale tool results, and [automatic compaction](https://platform.claude.com/docs/en/build-with-claude/compaction) with its threshold stops long loops from carrying their whole history forward.

8181

82The levers interact with the cache and each other, so judge them by net effect, and use [cache diagnostics](https://platform.claude.com/docs/en/build-with-claude/cache-diagnostics) to confirm your cached prefix survives each change. Anthropic turned the levers on one at a time for an issue-triage agent working through 20 real bug reports with screenshots from a public repository (and, for the second panel, a longer variant of the same job):

82Defer tool definitions you are not using. Every tool definition attached to a request is input on every turn, and a few MCP servers add up to hundreds of them. Anthropic ran the triage agent with its own two tools plus a catalog of real tool definitions from public MCP servers, for a total of up to 502 tools, loading all of them or marking the extras `defer_loading` behind [tool search](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-search-tool):

8383

84![Two bar charts of triage-run cost: caching cuts 83%, trimming reaches 88%; on the longer run, compaction cuts a further 38%](https://platform.claude.com/docs/images/cost-intel-hygiene.png)

84![Line chart: with all tools loaded, run cost rises from $0.55 to $1.02 at 502 tools; with tool search it stays at $0.56](https://platform.claude.com/docs/images/cost-intel-tool-search.png)

8585

86Caching did almost all the work, and trimming took the total to 88%. Each bar is one run, so differences of $0.10 are noise; those shown here are not. Compaction needs a session long enough to trigger it: the 20-issue run never reached the 50,000-token floor once its inputs were trimmed, but on the second panel's longer variant it fired once and cut the bill a further 38%.

86With every definition loaded, the run cost rose from $0.55 to $1.02, tracking the schema tokens on each request. With tool search, it stayed at $0.56 at every catalog size, 45% less at 502 tools. Accuracy was 15 to 18 of 20 in every cell either way, and the model never called a wrong tool, so at this scale the catalog costs money, not correctness. The same holds for tools that come through the [MCP connector](https://platform.claude.com/docs/en/agents-and-tools/mcp-connector): with a public GitHub MCP server attached, deferring its toolset (`default_config: {defer_loading: true}`) cut the run 20% at the same accuracy.

8787

88Context editing is the one lever here that is not free. Every clearing pass rewrites the cached conversation, which works against prompt caching; in this run, context editing cost more than it saved. Use it to make room in the context window, and [clear in a few large batches](https://platform.claude.com/docs/en/build-with-claude/context-editing#context-editing-and-prompt-caching).

88Keep data files out of the prompt. When the model has to compute over a table, upload it with the [Files API](https://platform.claude.com/docs/en/build-with-claude/files) and let the model query it with [code execution](https://platform.claude.com/docs/en/agents-and-tools/tool-use/code-execution-tool) instead of pasting it in. Anthropic asked 25 aggregate questions[15](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#refs) (sums, filtered counts, group-bys, and a date filter) over a 1,862-row public CSV, with the answers computed by pandas:

8989

90![Scatter chart: with the file uploaded and code execution, 25 of 25 correct at $0.40; pasted into the prompt, 6 of 25 at $5.01](https://platform.claude.com/docs/images/cost-intel-data-files.png)

92Pasted into the prompt, the table is about 91,000 input tokens on every request, and Claude Sonnet 5 answered 6 of 25 questions correctly at $5.01 per run. Uploaded, with code execution, it answered 25 of 25 at $0.40. Claude Opus 5 showed the same pattern (6 of 25 at $13.45 against 25 of 25 at $1.91).

94The levers interact with the cache and each other, so judge them by net effect, and use [cache diagnostics](https://platform.claude.com/docs/en/build-with-claude/cache-diagnostics) to confirm your cached prefix survives each change. Anthropic measured them on an issue-triage agent working through 20 real bug reports with screenshots from a public repository, and on a longer variant of the same job with 2.6 times the tokens. With caching on, input trimming (image resizing and tool search) took a further 26% off the short run and 21% off the long one. The context levers are where the two runs diverge:

96![Bar chart by run length: context editing adds 74% on the short run; compaction saves 32% and pruning 39% on the long](https://platform.claude.com/docs/images/cost-intel-hygiene.png)

98The context levers only pay on a session long enough to need them. On the 20-issue run, context editing cost 74% more, and compaction and the prune changed nothing. On the long run, context editing changed nothing, compaction saved 32%, and the prune saved 39%. The prune is a few lines you write yourself: at each task boundary, replace large stale tool results with a one-line extract. It caches well because the edits sit at the tail of the conversation, where the next task adds new content anyway: 89% cache reads on the first request after a boundary and 81% on the requests between boundaries. Run-wide, the prune and context editing cache equally well. The prune is cheaper because context editing rewrites content mid-task that the prune deletes (about two thirds of the gap) and because it keeps the context about half the size (the other third). If you use context editing, [clear in a few large batches](https://platform.claude.com/docs/en/build-with-claude/context-editing#context-editing-and-prompt-caching). The prune, adapted from the harness:

100```python

101import re

102

103PRUNED = "[pruned at issue boundary]"

104

105

106def prune_task_boundary(messages, tool_name_by_id, threshold=2000):

107 """Call once per task boundary. Replaces large, stale search results with a one-line extract."""

108 for message in messages:

109 if message["role"] != "user" or not isinstance(message["content"], list):

110 continue

111 for block in message["content"]:

112 if not (isinstance(block, dict) and block.get("type") == "tool_result"):

113 continue

114 if tool_name_by_id.get(block.get("tool_use_id")) != "search_issues":

115 continue

116 result_text = block.get("content")

117 if not isinstance(result_text, str) or len(result_text) <= threshold:

118 continue

119 if result_text.startswith(PRUNED):

120 continue # already pruned on an earlier boundary

121 # cap single-line results so the extract stays short

122 first_line = result_text.split("\n", 1)[0].strip()[:200]

123 refs = re.findall(r"#(\d+)", result_text)[:5]

124 extract = f"{PRUNED} {first_line}"

125 if refs:

126 extract += " kept refs: " + " ".join("#" + r for r in refs)

127 block["content"] = extract

128```

129

90130### Batch work that can wait

91131

92132The [Batch API](https://platform.claude.com/docs/en/build-with-claude/batch-processing) takes 50% off every token of a request, including cached ones, in exchange for results arriving any time within 24 hours. Route every request no one is waiting on through a batch, and keep the interactive path for the rest. Batching is the second-largest free lever after caching for unattended agent work: evaluation runs, backfills, and scheduled jobs such as a recurring run of the issue-triage agent from the [token-trimming measurement](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#trim-input-and-context-tokens). It combines with everything on this page except interactivity, but is not available for Claude Managed Agents sessions, which are interactive by design (see [Claude Managed Agents pricing](https://platform.claude.com/docs/en/about-claude/pricing#claude-managed-agents-pricing)).

from line 244

204244* **`max_tokens`** caps a single response, invisibly to the model, so lowering it does not make the model economize. The turns that needed the room are discarded and still billed. On an internal repository-task benchmark[12](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#refs), a 16,384-token cap ended 15% of Claude Opus 5's attempts and a third of Claude Fable 5's, none of them solved. Capped runs spent less per attempt but bought proportionally fewer solves, so cost per solved task was the same as at 64,000. At that setting nothing was cut off, and Fable solved 54.6% of tasks instead of 36.6% on the problems both runs scored (on a separate cut of the SWE-bench Pro[3](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#refs) subset, described in reference 12, 92% instead of 90%). Retrying capped attempts only adds cost: at the same cap they never succeeded, and at a higher one you also pay for the wasted attempt. Set `max_tokens` to 64,000 for agentic work (128,000, the maximum, at `xhigh` or `max` effort), [stream responses](https://platform.claude.com/docs/en/build-with-claude/streaming) that large, treat [`stop_reason: max_tokens`](https://platform.claude.com/docs/en/build-with-claude/handling-stop-reasons#max-tokens) as a failure, and save money with effort and task budgets, which the model can see.

205245* **Session budgets on Claude Managed Agents** are the hard stop. A [session budget](https://platform.claude.com/docs/en/managed-agents/budgets) is a dollar cap on one session at list rates for tokens, searches, and session time. At the cap, the session pauses with `stop_reason: budget_reached`; raising the budget resumes it. It is platform-enforced, works on any model with a list price (including Claude Sonnet 5), and combines with the advisory task budget. Deployments apply the same field to every run.

206246

247Ask for shorter answers. Output tokens cost five times input tokens on Claude Sonnet 5, and in an agent loop every token the model writes comes back as input on every later turn, so you pay for a long answer again and again. Anthropic ran the triage job under three final-answer instructions, three runs each, with the same model and tools. The original asked for two lines:

248

249```text wrap

2504. Finish with exactly two lines:

251LABEL: <one of: bug-confirmed, needs-more-info, duplicate-candidate, feature-request, upstream-issue, perf, ui-polish>

252SUMMARY: <one or two sentences for the engineering team>

253```

254

255The shorter variant asked for one:

256

257```text wrap

2584. Finish with exactly one line in this form:

259DECISION | LABEL | REASON

260where DECISION is one of: triage-now, needs-info, close-duplicate; LABEL is one of: bug-confirmed, needs-more-info, duplicate-candidate, feature-request, upstream-issue, perf, ui-polish; REASON is one clause under 15 words. Output nothing after that line.

261```

262

263The longer variant asked for a memo with five headed sections: problem summary, evidence, duplicate check, recommended label, and next steps. For one issue, a queued prompt that never sends after a skipped question, the first two answers were:

264

265```text wrap

266LABEL: bug-confirmed

267SUMMARY: When a user submits a new prompt instead of answering an agent's pending question, the question is cancelled/skipped but the new prompt remains stuck in "QUEUED" state indefinitely since it's waiting on a response to the now-cancelled question; the queued prompt should be processed immediately after cancellation.

268```

269

270```text wrap

271triage-now | bug-confirmed | Clear repro steps show prompt queues indefinitely after cancelled question.

272```

273

274![Bar chart: one-line format $0.49 per run, original two-line format $0.57, memo $1.40, all 78% to 85% correct](https://platform.claude.com/docs/images/cost-intel-output-format.png)

275

276The one-line answer used 39% fewer output tokens and cost $0.49 for the run against $0.57, with the same accuracy against the gold labels (78% compared with 80%, inside run-to-run noise). The memo used six times the output tokens of the original and cost $1.40, 2.8 times the one-line answer, for 85%, also inside the noise. The three formats are equally accurate; they differ in what you pay. Ask for the answer you will read, not the one that looks thorough.

277

207278The first of two `max_tokens` charts plots cost per attempt and per solved task at each cap:

208279

209280![Bar charts: at a 16k cap both models spend less per attempt but the same per solved task as at 64k, as they solve fewer tasks](https://platform.claude.com/docs/images/cost-intel-max-tokens-saving.png)

from line 306

235306

236307The first is the gap between the models. The advisor can only hand over capability the executor lacks: on GPQA Diamond[9](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#refs) a Claude Haiku 4.5 executor gained a great deal from a Claude Opus 5 advisor, a Claude Sonnet 5 executor gained a few points, and a frontier executor almost nothing.

237308

238The second, and the fragile one, is whether the executor actually asks (the consult rate). An executor at low effort can stop noticing it is stuck: a pairing that consults on most tasks at the default effort can fall to consulting on almost none when effort is lowered, and then scores below the executor alone. The rate also varies by task: on DeepSWE[10](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#refs) a low-effort Sonnet 5 executor kept asking and gained 23 points; on SWE-bench Pro[3](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#refs) the same executor stopped. When the executor does ask, it gets most of the way there. Across the pairings in the following chart, the advisor closed 60% to 90% of the gap to the stronger model while that model was paid for only on the consultations, which makes the cost cases possible:

309The second, and the fragile one, is whether the executor actually asks (the consult rate). An executor at low effort can stop detecting that it is stuck: a pairing that consults on most tasks at the default effort can fall to consulting on almost none when effort is lowered, and then scores below the executor alone. The rate also varies by task: on DeepSWE[10](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#refs) a low-effort Sonnet 5 executor kept asking and gained 23 points; on SWE-bench Pro[3](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#refs) the same executor stopped. When the executor does ask, it recovers most of the gap. Across the pairings in the following chart, the advisor closed 50% to 90% of the gap to the stronger model and you pay for the stronger model only on the consultations, which is what makes the cost cases possible:

239310

240311![Bar chart of six advisor pairings: gap available versus gain realized, labeled with consult rates, which the gains track](https://platform.claude.com/docs/images/cost-intel-advisor-mechanism.png)

241312

from line 599

528599

529600The following table lists the levers in the order to try them:

530601

532| -------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------ | ------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |

537| Prompt audit against the current model | 14% on both migrations measured | None; a gain on one | Faster (fewer tool rounds) | [Audit prompts against the current model](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#audit-prompts-against-the-current-model) |

538| Lower effort | Knowledge work: `medium` 15% to 30%, `low` a third to a half; long coding: `medium` about half, `low` about three quarters | 1 to 3 points on knowledge work, 2 to 8 on long coding | Faster | [Tune effort](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#tune-effort) |

539| Re-run failures | About half, at the same pass rate | None | Two runs on the tasks that fail | [Re-run failures at higher effort](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#re-run-failures-at-higher-effort) |

541| Raising `max_tokens` | None per solved task, but more tasks solved | Gains of 2 to 18 points | Neutral | [Set budgets and output caps](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#set-budgets-and-output-caps) |

542| Advisor | Depends on the capability gap and the consult rate; the chart-reading pairing scored above both models' effort curves, the coding pairing only marginally | Small gains | About two extra calls per task | [Advisor strategy](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#advisor-strategy-escalate-hard-decisions) |

543| Orchestrator | More than 60% below the frontier model beyond one context window; about half on routine tails | 2 to 6 points below the frontier model | Much faster on large inputs | [Orchestrator strategy](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#orchestrator-strategy-delegate-bulk-work) |

603| ------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------ | ------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |

606| Prune stale tool results at task boundaries | 39% on the long triage run (compaction 32%); nothing on short loops | None measured | Neutral | [Trim input and context tokens](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#trim-input-and-context-tokens) |

607| Tool search | 45% with 500 tool definitions attached; 20% with a GitHub MCP server | None | Neutral | [Trim input and context tokens](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#trim-input-and-context-tokens) |

608| Data files through code execution | 92% on a 25-question data task | A gain, 25 of 25 instead of 6 of 25 | Faster | [Trim input and context tokens](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#trim-input-and-context-tokens) |

610| Prompt audit against the current model | 14% on both migrations measured | None; a gain on one | Faster (fewer tool rounds) | [Audit prompts against the current model](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#audit-prompts-against-the-current-model) |

611| Lower effort | Knowledge work: `medium` 15% to 30%, `low` a third to a half; long coding: `medium` about half, `low` about three quarters | 1 to 3 points on knowledge work, 2 to 8 on long coding | Faster | [Tune effort](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#tune-effort) |

612| Re-run failures | About half, at the same pass rate | None | Two runs on the tasks that fail | [Re-run failures at higher effort](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#re-run-failures-at-higher-effort) |

615| Raising `max_tokens` | None per solved task, but more tasks solved | Gains of 2 to 18 points | Neutral | [Set budgets and output caps](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#set-budgets-and-output-caps) |

616| Advisor | Depends on the capability gap and the consult rate; the chart-reading pairing scored above both models' effort curves, the coding pairing only marginally | Small gains | About two extra calls per task | [Advisor strategy](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#advisor-strategy-escalate-hard-decisions) |

617| Orchestrator | More than 60% below the frontier model beyond one context window; about half on routine tails | 2 to 6 points below the frontier model | Much faster on large inputs | [Orchestrator strategy](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#orchestrator-strategy-delegate-bulk-work) |

544618

545619## Benchmarks referenced

546620

547621All measurements are Anthropic-internal runs of these benchmarks. Unless noted, costs are USD at August 2026 list prices; Claude Sonnet 5 figures use $2 and $10 per million input and output tokens. Charts labeled "notional USD" price each request's token counts at those rates rather than reporting invoices.

548622

5491. **WideSearch:** Wong et al., "WideSearch: Benchmarking Agentic Broad Info-Seeking," arXiv.07999, 2025. Broad web-research tasks graded on a many-row table's completeness and accuracy; 200 problems, 3 runs per configuration. The caching and effort charts come from separate runs, so per-problem costs differ slightly. The cost-concentration chart is a separate 20-problem run, 3 runs per problem, costed from per-request billing records.

6231. **WideSearch:** Wong et al., "WideSearch: Benchmarking Agentic Broad Info-Seeking," arXiv:2508.07999, 2025. Broad web-research tasks graded on a many-row table's completeness and accuracy; 200 problems, 3 runs per configuration. The caching and effort charts come from separate runs, so per-problem costs differ slightly. The cost-concentration chart is a separate 20-problem run, 3 runs per problem, costed from per-request billing records.

5506242. **GDPval:** OpenAI, "GDPval: Evaluating AI Model Performance on Real-World Economically Valuable Tasks," 2025. Knowledge-work deliverables graded against task rubrics; a 210-task run of the released gold set, one attempt per task. A Claude model grades, so absolute scores may differ from published results.

5513. **SWE-bench Pro:** Scale AI, "SWE-Bench Pro: Can AI Agents Solve Long-Horizon Software Engineering Tasks?", 2025. A 482-problem subset selected for compatibility with Anthropic's evaluation harness; scores are not comparable to the public leaderboard. Claude Opus 5 at the default effort averages two runs; reduced-effort settings are single runs. Escalation figures come task by task from those runs: `low` first, then the default on its failures, solved 92.5% to 93.6% across run pairings for about $0.70; `medium` first, 93.8% to 94.2% for about $0.95; the default re-run on its own failures, 94.0% for $1.58; everything at the default, 90.9% to 92.5% for $1.39. The Claude Sonnet 5 executor pairings on the advisor chart come from the same August 2026 series on this subset: the Sonnet-plus-Opus pairing was run twice (a run and an exact replication) and the low-effort pairing once; the task-budget figures are one run per budget on the same subset. The Claude Fable 5 figure in [Compare models](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#compare-models-on-cost-per-task) is a single July 2026 run, also the task-budget chart's unbudgeted baseline; every budgeted run completed all 482 problems without harness errors.

6253. **SWE-bench Pro:** Scale AI, "SWE-Bench Pro: Can AI Agents Solve Long-Horizon Software Engineering Tasks?", 2025. A 482-problem subset selected for compatibility with Anthropic's evaluation harness; scores are not comparable to the public leaderboard. Claude Opus 5 at the default effort averages two runs; reduced-effort settings are single runs. Escalation figures come task by task from those runs: `low` first, then the default on its failures, solved 92.5% to 93.6% across run pairings for about $0.70; `medium` first, 93.8% to 94.2% for about $0.95; the default re-run on its own failures, 94.0% for $1.58; everything at the default, 90.9% to 92.5% for $1.39. The Claude Sonnet 5 executor pairings on the advisor chart come from the same August 2026 series on this subset: the Sonnet-plus-Opus pairing was run twice (a run and an exact replication), the low-effort pairing once, and Claude Sonnet 5 alone twice (77.4%, the baseline for both Pro rows); the task-budget figures are one run per budget on the same subset. The Claude Fable 5 figure in [Compare models](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#compare-models-on-cost-per-task) is a single July 2026 run, also the task-budget chart's unbudgeted baseline; every budgeted run completed all 482 problems without harness errors.

5526264. **BrowseComp:** Wei et al., "BrowseComp: A Simple Yet Challenging Benchmark for Browsing Agents," OpenAI, 2025. Effort figures use a 500-problem cut, one to three runs per setting. The cost-insurance chart uses 10 reliably solved problems from a 26-problem slice, 50 delegated and 70 solo runs ($6.45 compared with $11.99 per run in expectation); delegated figures carry a measurement band of about 20%.

5535. **Agent-architecture scaling:** Kim et al., "Towards a Science of Scaling Agent Systems," arXiv.08296, 2025. Independent external study, cited only for the direction of the finding on when delegation does not pay, not for any figure.

5546. **DeepWideSearch:** "DeepWideSearch: Benchmarking Depth and Width in Agentic Information Seeking," arXiv.20168, 2025. The 220 questions span 15 domains, each combining many-row collection with multi-hop retrieval; measured on the benchmark's standing row set, 3 runs per configuration.

5557. **DeepResearch Bench II:** Li et al., "DeepResearch Bench II: Diagnosing Deep Research Agents via Rubrics from Expert Report," arXiv.08536, 2026. Its 132 research tasks across 22 domains are graded against expert-derived binary rubrics; measured on a 50-task subset stratified across all themes, one attempt per task, 3 runs, scored on tasks no configuration refused. Claude Opus 4.6 judges under the benchmark's rubric protocol; the original uses a different judge, and an Anthropic judge may favor the house style. The runs predate Claude Opus 5, hence its absence from the [Compare models](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#compare-models-on-cost-per-task) chart. The Sonnet 5 cost differs slightly between the caching chart (inference cost, with and without caching) and that chart (all-in cost, caching on); both come from the same runs.

6275. **Agent-architecture scaling:** Kim et al., "Towards a Science of Scaling Agent Systems," arXiv:2512.08296, 2025. Independent external study, cited only for the direction of the finding on when delegation does not pay, not for any figure.

6286. **DeepWideSearch:** "DeepWideSearch: Benchmarking Depth and Width in Agentic Information Seeking," arXiv:2510.20168, 2025. The 220 questions span 15 domains, each combining many-row collection with multi-hop retrieval; measured on the benchmark's standing row set, 3 runs per configuration.

6297. **DeepResearch Bench II:** Li et al., "DeepResearch Bench II: Diagnosing Deep Research Agents via Rubrics from Expert Report," arXiv:2601.08536, 2026. Its 132 research tasks across 22 domains are graded against expert-derived binary rubrics; measured on a 50-task subset stratified across all themes, one attempt per task, 3 runs, scored on tasks no configuration refused. Claude Opus 4.6 judges under the benchmark's rubric protocol; the original uses a different judge, and an Anthropic judge may favor the house style. The runs predate Claude Opus 5, hence its absence from the [Compare models](https://platform.claude.com/docs/en/about-claude/models/optimizing-for-cost-and-intelligence#compare-models-on-cost-per-task) chart. The Sonnet 5 cost differs slightly between the caching chart (inference cost, with and without caching) and that chart (all-in cost, caching on); both come from the same runs.

5566308. **Corpus defect sweep:** Anthropic-internal, for work larger than one context window: a 21.6-million-token corpus from 14 public Python package sources with 130 planted defects and deterministic grading; protocol fixed before the runs and internally reviewed; three runs per configuration. Every configuration ran on Claude Managed Agents. The charted team configuration is an August 2026 run in which the Claude Fable 5 coordinator ran the whole sweep inside the platform at its documented limit of 25 concurrent Claude Sonnet 5 workers; its three episodes scored F1 0.842, 0.805, and 0.810 for $263, $299, and $261. The solo configurations are July 2026 runs on the same corpus build. Absolute F1 is specific to this corpus build, not comparable across benchmarks; configuration comparisons are like for like.

5576319. **GPQA Diamond:** Rein et al., "GPQA: A Graduate-Level Google-Proof Q\&A Benchmark," 2023. The 198-question Diamond subset, measured August 2026, two runs per configuration, model-graded against reference answers, advisor tokens metered per request. A platform safety check refused two biology questions on the Sonnet and Opus executors; excluding them changes no comparison by more than one point.

55810. **DeepSWE:** Datacurve, "DeepSWE: Measuring Frontier Coding Agents on Original, Long-Horizon Engineering Tasks," arXiv.07946, 2026. Measured August 2026: 113 original tasks across five languages with program-based verifiers. Pairings are two runs each with advisor tokens metered per request, and used a client-side advisor loop rather than the advisor tool, with identical accounting. Single-model effort sweeps are single runs priced from token counts, a cache-aware approximation. Costs per task are run totals divided by 113.

63210. **DeepSWE:** Datacurve, "DeepSWE: Measuring Frontier Coding Agents on Original, Long-Horizon Engineering Tasks," arXiv:2607.07946, 2026. Measured August 2026: 113 original tasks across five languages with program-based verifiers. Pairings are two runs each with advisor tokens metered per request, and used a client-side advisor loop rather than the advisor tool, with identical accounting. Single-model effort sweeps are single runs priced from token counts, a cache-aware approximation. Costs per task are run totals divided by 113.

55963311. **Internal agentic-coding benchmark:** Anthropic-internal: 370 repository tasks graded by the repositories' own tests. The API figures (Opus 5 alone, Fable 5 alone, and the pairing) were measured August 2026 at the default effort with a 128,000-token output cap, one run per configuration: five attempts per task at the default settings and for the pairing, one at `low` and `medium`; the pairing averaged about two advisor consultations per attempt; costs are per attempt. The Claude Code figures are July 2026 runs of the same tasks, one run per configuration, costs approximate.

56063412. **Internal repository-task benchmark (cap measurement):** A separate Anthropic-internal set of about 130 repository tasks, run August 2026 with a plain API agent loop, one attempt per task. The 16,384-token figures average two runs per model; the 64,000-token figures are single runs (124 tasks scored for Opus 5; 108 for Claude Fable 5, the environment having skipped the rest before the model ran). About half the Fable attempts the 16,384 cap had ended solved at 64,000; a further Fable run at 128,000 scored 56.1%, within noise of the 64,000 run. The SWE-bench Pro cap figures are one Claude Fable 5 run per cap at the default effort on a 100-problem subset stratified from reference 3's 482-problem set, not comparable to its scores. The chart's per-turn distributions come from the Opus run at 64,000 and the Fable run at 128,000, so neither is cut off by its own cap.

56163513. **Chartography:** Surge AI, "Chartography," 2026. The complete released 100-question set, measured August 2026 with Anthropic's implementation on Claude Managed Agents (standard cloud sandbox; advisor configurations use the Managed Agents advisor). Claude Sonnet 4.6 grades instead of the reference judge and the benchmark runs with tools, so scores compare across configurations here but not to the published leaderboard. Two runs per configuration, pooled; run-to-run spreads were 4 to 10 points. Costs exclude sandbox time, which added under 1%. The consult-rate comparison comes from rerunning the same configurations on the Messages API with a container tool set.

56263614. **Support-desk prompt-audit evaluation:** An Anthropic-constructed set of 44 support tickets with deterministic grading, run August 2026 under six system prompts, each adding to the same clean prompt one pattern common in prompts written for Claude Opus 4.8 and Claude Sonnet 4.6. Each chart point is one of three cases (older model, newer model on the same prompt, newer model after the audit) averaged over the six prompts and 44 tickets. The Opus 5 accuracy gain has a 95% confidence interval of 3 to 8 points; the Sonnet accuracy differences are within noise.

63715. **Data-file question set:** An Anthropic-constructed set of 25 aggregate questions over a 1,862-row slice of a public liquor-sales CSV, with ground truth computed by pandas and exact-match grading, run August 2026 on Claude Sonnet 5 and Claude Opus 5 with thinking disabled (the in-context arm cannot complete at the default), a 4,000-token output cap, and no prompt caching, three runs per configuration. The file arm uploads the CSV through the Files API and uses the `code_execution_20260120` tool.

563638

564639## Next steps

565640

about-claude/models/whats-new-opus-5 Changed · +3 / -3 lines

from line 32

3232

3333### Fast mode

3434

35[Fast mode](https://platform.claude.com/docs/en/build-with-claude/fast-mode) (research preview) is available for Claude Opus 5 on the Claude API only; it is not currently available on Amazon Bedrock, Google Cloud, or Microsoft Foundry. Fast mode for Claude Opus 5 is priced at $10 USD per million input tokens and $50 USD per million output tokens. See [Fast mode](https://platform.claude.com/docs/en/build-with-claude/fast-mode) for access, supported models, and pricing.

35[Fast mode](https://platform.claude.com/docs/en/build-with-claude/fast-mode) (research preview) is available for Claude Opus 5 on the Claude API only; it is not currently available on Amazon Bedrock, Claude Platform on AWS, Google Cloud, or Microsoft Foundry. Fast mode for Claude Opus 5 is priced at $10 USD per million input tokens and $50 USD per million output tokens. See [Fast mode](https://platform.claude.com/docs/en/build-with-claude/fast-mode) for access, supported models, and pricing.

3636

3737## Behavior changes

3838

from line 234

234234

235235### Disabling thinking requires effort `high` or below

236236

237On Claude Opus 5, `thinking: {"type": "disabled"}` is accepted only when the effort level is `high` or below. Setting `thinking: {"type": "disabled"}` with effort `xhigh` or `max` returns a 400 error. This is generally available behavior on Claude Opus 5 onward, enforced on each request, and it is a breaking change from Claude Opus 4.8, where disabling thinking was independent of the effort level. If you disable thinking at high effort levels today, either keep thinking disabled and set effort to `high` or below, or keep the effort level and remove the `thinking` field.

237On Claude Opus 5, `thinking: {"type": "disabled"}` is accepted only when the effort level is `high` or below. Setting `thinking: {"type": "disabled"}` with effort `xhigh` or `max` returns a 400 error. This rule is enforced on every request to Claude Opus 5 and later models. It is a breaking change from Claude Opus 4.8, where disabling thinking was independent of the effort level. If you disable thinking at high effort levels today, either keep thinking disabled and set effort to `high` or below, or keep the effort level and remove the `thinking` field.

238238

239239With thinking disabled, Claude Opus 5 can occasionally write a tool call into its text output instead of emitting a `tool_use` block, or include internal XML tags in its visible response. Where possible, keep thinking enabled and control token cost with lower effort levels; for integrations that must keep thinking disabled, see [Running with thinking disabled](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-opus-5#running-with-thinking-disabled) for prompting mitigations.

240240

from line 269

269269Claude Opus 5 is available on:

270270

271271* **Claude API:** available to all customers, as `claude-opus-5`.

272* **AWS:** available through [Claude in Amazon Bedrock](https://platform.claude.com/docs/en/build-with-claude/claude-in-amazon-bedrock), as `anthropic.claude-opus-5`. Claude Opus 5 is also reachable through the `InvokeModel` API on `bedrock-runtime`, served by the same infrastructure; the [Claude on Amazon Bedrock (legacy)](https://platform.claude.com/docs/en/build-with-claude/claude-on-amazon-bedrock-legacy) integration does not include it in its ARN-versioned model ID table.

272* **AWS:** available through [Claude in Amazon Bedrock](https://platform.claude.com/docs/en/build-with-claude/claude-in-amazon-bedrock), as `anthropic.claude-opus-5`, and through [Claude Platform on AWS](https://platform.claude.com/docs/en/build-with-claude/claude-platform-on-aws). On Amazon Bedrock, Claude Opus 5 is also reachable through the `InvokeModel` API on `bedrock-runtime`, served by the same infrastructure; the [Claude on Amazon Bedrock (legacy)](https://platform.claude.com/docs/en/build-with-claude/claude-on-amazon-bedrock-legacy) integration does not include it in its ARN-versioned model ID table.

273273* **Google Cloud:** available through [Claude on Google Cloud](https://platform.claude.com/docs/en/build-with-claude/claude-on-vertex-ai), as `claude-opus-5`.

274274* **Microsoft Foundry:** available through [Claude in Microsoft Foundry](https://platform.claude.com/docs/en/build-with-claude/claude-in-microsoft-foundry).

275275

agents-and-tools/agent-skills/quickstart Changed · +16 / -26 lines

from line 95

9595 ```

9696

9797 ```php PHP

98 // The PHP SDK exposes the Skills API under the beta namespace; field names can differ from other SDKs.

9998 // List Anthropic-managed Skills

100 $skills = $client->beta->skills->list(source: 'anthropic');

99 $skills = $client->skills->list(source: 'anthropic');

101100

102 foreach ($skills->data as $skill) {

103 echo "{$skill->id}: {$skill->displayTitle}\n";

101 foreach ($skills->getItems() as $skill) {

102 echo "{$skill->id}: {$skill->displayName}\n";

104103 }

105104 ```

106105

from line 300

301300 ```

302301

303302 ```php PHP

304 // The PHP SDK supports container skills only through $client->beta->messages with the skills beta.

305303 // Create a message with the PowerPoint Skill

306 $response = $client->beta->messages->create(

304 $response = $client->messages->create(

307305 model: 'claude-opus-5',

308306 maxTokens: 16000,

309 betas: ['skills-2025-10-02'],

310307 container: [

311 'skills' => [['type' => 'anthropic', 'skill_id' => 'pptx', 'version' => 'latest']],

308 'skills' => [['type' => 'anthropic', 'skillID' => 'pptx', 'version' => 'latest']],

312309 ],

313310 messages: [

314311 [

from line 543

546543 ```

547544

548545 ```php PHP

549 // The PHP SDK exposes the Files API under the beta namespace; field names can differ from other SDKs.

550546 // Extract the file ID. The code execution tool runs the Skill's code through

551547 // its Bash sub-tool, and generated files appear as bash_code_execution_output

552548 // items inside the bash_code_execution_tool_result block.

from line 563

567563 if ($fileId !== null) {

568564 // Download the file and save it

569565 $outputPath = sys_get_temp_dir() . '/renewable_energy.pptx';

570 $fileContent = $client->beta->files->download($fileId);

566 $fileContent = $client->files->download($fileId);

571567 file_put_contents($outputPath, $fileContent);

572568 echo "Presentation saved to {$outputPath}\n";

573569 }

from line 757

761757 ```

762758

763759 ```php PHP

764 // The PHP SDK supports container skills only through $client->beta->messages with the skills beta.

765 $response = $client->beta->messages->create(

760 $response = $client->messages->create(

766761 model: 'claude-opus-5',

767762 maxTokens: 16000,

768 betas: ['skills-2025-10-02'],

769763 container: [

770764 'skills' => [

771 ['type' => 'anthropic', 'skill_id' => 'xlsx', 'version' => 'latest'],

765 ['type' => 'anthropic', 'skillID' => 'xlsx', 'version' => 'latest'],

772766 ],

773767 ],

774768 messages: [

from line 771

777771 'content' => 'Create a quarterly sales tracking spreadsheet with sample data',

778772 ],

779773 ],

780 tools: [new BetaCodeExecutionTool20260521()],

774 tools: [['type' => 'code_execution_20260521', 'name' => 'code_execution']],

781775 );

782776 ```

783777

from line 950

956950 ```

957951

958952 ```php PHP

959 // The PHP SDK supports container skills only through $client->beta->messages with the skills beta.

960 $response = $client->beta->messages->create(

953 $response = $client->messages->create(

961954 model: 'claude-opus-5',

962955 maxTokens: 16000,

963 betas: ['skills-2025-10-02'],

964956 container: [

965957 'skills' => [

966 ['type' => 'anthropic', 'skill_id' => 'docx', 'version' => 'latest'],

958 ['type' => 'anthropic', 'skillID' => 'docx', 'version' => 'latest'],

967959 ],

968960 ],

969961 messages: [

from line 964

972964 'content' => 'Write a 2-page report on the benefits of renewable energy',

973965 ],

974966 ],

975 tools: [new BetaCodeExecutionTool20260521()],

967 tools: [['type' => 'code_execution_20260521', 'name' => 'code_execution']],

976968 );

977969 ```

978970

from line 1143

11511143 ```

11521144

11531145 ```php PHP

1154 // The PHP SDK supports container skills only through $client->beta->messages with the skills beta.

1155 $response = $client->beta->messages->create(

1146 $response = $client->messages->create(

11561147 model: 'claude-opus-5',

11571148 maxTokens: 16000,

1158 betas: ['skills-2025-10-02'],

11591149 container: [

11601150 'skills' => [

1161 ['type' => 'anthropic', 'skill_id' => 'pdf', 'version' => 'latest'],

1151 ['type' => 'anthropic', 'skillID' => 'pdf', 'version' => 'latest'],

11621152 ],

11631153 ],

11641154 messages: [

from line 1157

11671157 'content' => 'Generate a PDF invoice template',

11681158 ],

11691159 ],

1170 tools: [new BetaCodeExecutionTool20260521()],

1160 tools: [['type' => 'code_execution_20260521', 'name' => 'code_execution']],

11711161 );

11721162 ```

11731163

agents-and-tools/tool-use/browser-use-tool Changed · +2 / -3 lines

from line 17

1717

1818## Quick start

1919

20The browser use tool is generally available on the Claude API with no beta header: add one entry of type `browser_toolset_20260801`, with no `name`, to the `tools` array of a [Messages API](https://platform.claude.com/docs/en/api/messages/create) request.

20The browser use tool is available on the Claude API with no beta header: add one entry of type `browser_toolset_20260801`, with no `name`, to the `tools` array of a [Messages API](https://platform.claude.com/docs/en/api/messages/create) request.

2121

2222<CodeGroup>

2323 ```bash cURL

from line 976

976976 $failed = false;

977977 foreach ($response->content as $block) {

978978 // This example declares only the browser toolset; route other tools here if you add them.

979 // Read toolset_name through array access: the SDK keeps it as raw data until a release types it.

980 if (!($block instanceof ToolUseBlock) || ($block['toolsetName'] ?? $block['toolset_name'] ?? null) !== 'browser') {

979 if (!($block instanceof ToolUseBlock) || $block->toolsetName !== 'browser') {

981980 continue;

982981 }

983982 $result = ['type' => 'tool_result', 'tool_use_id' => $block->id, 'toolset_name' => 'browser'];

agents-and-tools/tool-use/code-execution-tool Changed · +12 / -18 lines

from line 42

4242* `code_execution_20260120` adds REPL state persistence and [programmatic tool calling](https://platform.claude.com/docs/en/agents-and-tools/tool-use/programmatic-tool-calling) from within the sandbox. Claude Haiku 4.5 accepts the `code_execution_20260120` and `code_execution_20260521` tool types, but programmatic tool calling and the REPL state persistence that depends on it aren't available on it, so the newer versions behave like `code_execution_20250825` there.

4343* `code_execution_20260521` is the same runtime as `code_execution_20260120`. The difference is that the tool description tells Claude about the 90-second wall-clock limit on each Python cell in programmatic tool calling, so Claude can budget long-running cells. A cell that exceeds the limit returns a normal code execution result with a non-zero `return_code` and a `detection_timeout` status message in its output. This is separate from the `execution_time_exceeded` [error code](https://platform.claude.com/docs/en/agents-and-tools/tool-use/code-execution-tool#errors), which the API returns when a whole tool invocation exceeds the maximum execution time.

4444

45All three tool versions are generally available and don't require an `anthropic-beta` header. The legacy code execution beta headers remain valid opt-ins.

45None of the three tool versions requires an `anthropic-beta` header. The legacy code execution beta headers remain valid opt-ins.

4646

4747The examples on this page use `code_execution_20250825`, which covers the Bash and file operations they demonstrate and behaves the same way on every model in the table; use `code_execution_20260120` or later when you need programmatic tool calling or REPL state persistence. The current [web search](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-search-tool) and [web fetch](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-fetch-tool) tools (`web_search_20260209`, `web_fetch_20260209`, and later) require `code_execution_20260120` or later as their code execution version.

4848

from line 324

324324

325325 ```bash CLI

326326 # First, upload a file and capture the file ID

327 FILE_ID=$(ant files upload \

328 --file ./data.csv \

329 --transform id --raw-output)

327 FILE_ID=$(ant files upload --file ./data.csv --transform id --raw-output)

330328

331329 # Then use the file_id with code execution

332330 ant messages create <<YAML

from line 503

505503 ```

506504

507505 ```php PHP

508 // The PHP SDK exposes the Files API under the beta namespace; field names can differ from other SDKs.

509506 $client = new Client();

510507

511508 // Upload a file

512 $fileObject = $client->beta->files->upload(

509 $fileObject = $client->files->upload(

513510 file: FileParam::fromResource(fopen('data.csv', 'r')),

514511 );

515512

516513 // Use the file_id with code execution

517 $response = $client->beta->messages->create(

514 $response = $client->messages->create(

518515 model: Model::CLAUDE_OPUS_5,

519516 maxTokens: 4096,

520 betas: [AnthropicBeta::FILES_API_2025_04_14],

521517 messages: [

522518 [

523519 'role' => 'user',

524520 'content' => [

525 BetaTextBlockParam::with(text: 'Analyze this CSV data'),

526 BetaContainerUploadBlockParam::with(fileID: $fileObject->id),

521 TextBlockParam::with(text: 'Analyze this CSV data'),

522 ContainerUploadBlockParam::with(fileID: $fileObject->id),

527523 ],

528524 ],

529525 ],

530 tools: [new BetaCodeExecutionTool20250825()],

526 tools: [new CodeExecutionTool20250825()],

531527 );

532528

533529 echo json_encode($response), PHP_EOL;

from line 803

807803 ```

808804

809805 ```php PHP

810 // The PHP SDK exposes the Files API under the beta namespace; field names can differ from other SDKs.

811806 $client = new Client();

812807

813808 // Request code execution that creates files

814 $response = $client->beta->messages->create(

809 $response = $client->messages->create(

815810 maxTokens: 4096,

816811 messages: [

817812 [

from line 815

820815 ],

821816 ],

822817 model: Model::CLAUDE_OPUS_5,

823 betas: [AnthropicBeta::FILES_API_2025_04_14],

824 tools: [new BetaCodeExecutionTool20250825()],

818 tools: [new CodeExecutionTool20250825()],

825819 );

826820

827821 /**

from line 823

829823 *

830824 * @return list<string>

831825 */

832 function extractFileIds(BetaMessage $response): array

826 function extractFileIds(Message $response): array

833827 {

834828 $fileIds = [];

835829 foreach ($response->content as $block) {

from line 843

849843

850844 // Download the created files

851845 foreach (extractFileIds($response) as $fileId) {

852 $fileMetadata = $client->beta->files->retrieveMetadata($fileId);

853 $fileContent = $client->beta->files->download($fileId);

846 $fileMetadata = $client->files->retrieveMetadata($fileId);

847 $fileContent = $client->files->download($fileId);

854848

855849 file_put_contents($fileMetadata->filename, $fileContent);

856850 echo "Downloaded: {$fileMetadata->filename}\n";

agents-and-tools/tool-use/computer-use-tool Changed · +2 / -3 lines

from line 18

1818For tasks that stay inside webpages, the [browser use tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/browser-use-tool) is the closer fit: its member tools read and act on the page itself, and it doesn't need a full desktop environment.

1919

2020<Note>

21 Computer use is generally available on the Claude API as the `computer_toolset_20260801` toolset, with no beta header; see [Compatibility](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool#compatibility) for the supported models.

21 Computer use is available on the Claude API as the `computer_toolset_20260801` toolset, with no beta header; see [Compatibility](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool#compatibility) for the supported models.

2222

2323 Existing `computer_20251124` integrations keep working, and earlier tool versions remain available in beta for models and platforms that don't support the toolset. See [Migrate from `computer_20251124`](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool#migrate-from-computer-20251124) to upgrade, or [Earlier tool versions](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool#earlier-tool-versions) for the beta headers.

2424</Note>

from line 1495

14951495 $failed = false;

14961496 foreach ($response->content as $block) {

14971497 // This example declares only the computer toolset; route other tools here if you add them.

1498 // Read toolset_name through array access: the SDK keeps it as raw data until a release types it.

1499 if (!($block instanceof ToolUseBlock) || ($block['toolsetName'] ?? $block['toolset_name'] ?? null) !== 'computer') {

1498 if (!($block instanceof ToolUseBlock) || $block->toolsetName !== 'computer') {

15001499 continue;

15011500 }

15021501 $result = ['type' => 'tool_result', 'tool_use_id' => $block->id, 'toolset_name' => 'computer'];

agents-and-tools/tool-use/tool-reference Changed · +14 / -14 lines

from line 10

1010

1111Anthropic provides two kinds of tools: **server tools** that execute on Anthropic's infrastructure, and **client tools** where Anthropic defines the schema but your application handles execution. Both kinds appear in your request's `tools` array alongside any user-defined tools.

1212

14| -------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------- | --------- | ------------------------------------------------------------------ |

24| [Computer use tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool) | `computer_toolset_20260801` `computer_20251124` `computer_20250124` | Client | GA Beta: `computer-use-2025-11-24` Beta: `computer-use-2025-01-24` |

24| [Computer use tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool) | `computer_toolset_20260801` `computer_20251124` `computer_20250124` | Client | None `computer-use-2025-11-24` `computer-use-2025-01-24` |

2626

2727For model compatibility, see each tool's page. Supported models vary by tool and by tool version.

2828

from line 40

4040* **Model-keyed:** `text_editor_20250728` is for Claude 4 and later models and `text_editor_20250124` is for earlier models. The version you use depends on the model you target.

4141* **Variant, not version:** `tool_search_tool_regex_20251119` and `tool_search_tool_bm25_20251119` are two search algorithms released together. Neither supersedes the other.

4242* **Legacy:** `code_execution_20250522` supports only Python. `code_execution_20250825` adds Bash and file operations.

43* **Successor:** `computer_toolset_20260801` is the generally available successor to the beta `computer_20251124` and `computer_20250124` versions, which remain available for existing integrations and for models that don't support the toolset ([Earlier tool versions](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool#earlier-tool-versions)). `browser_toolset_20260801` is the first version of the browser use tool. Both are [client toolsets](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-reference#client-toolsets).

43* **Successor:** `computer_toolset_20260801` is the stable successor to the beta `computer_20251124` and `computer_20250124` versions, which remain available for existing integrations and for models that don't support the toolset ([Earlier tool versions](https://platform.claude.com/docs/en/agents-and-tools/tool-use/computer-use-tool#earlier-tool-versions)). `browser_toolset_20260801` is the first version of the browser use tool. Both are [client toolsets](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-reference#client-toolsets).

4444

4545The `mcp_toolset` type is not date-versioned; versioning is carried in the `anthropic-beta` header instead.

4646

agents-and-tools/tool-use/tool-search-tool Changed · +5 / -5 lines

from line 11

1111* **Context bloat:** A typical multiserver setup (GitHub, Slack, Sentry, Grafana, and Splunk) can consume \~55k tokens in definitions before Claude does any work. Tool search typically reduces this by over 85 percent, loading only the 3–5 tools Claude needs for a given request.

1212* **Tool selection accuracy:** Claude's ability to pick the right tool degrades once you exceed 30–50 available tools. Because tool search loads only a focused set of relevant tools on demand, selection accuracy stays high even across thousands of tools.

1313

14Tool search is generally available on the Claude API. For supported models, see [Model compatibility](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-search-tool#model-compatibility).

14For the models that support tool search, see [Model compatibility](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-search-tool#model-compatibility).

1515

1616<Tip>

1717 For background on the scaling challenges that tool search solves, see [Advanced tool use](https://www.anthropic.com/engineering/advanced-tool-use). Tool search's on-demand loading is also an instance of the broader just-in-time retrieval principle described in [Effective context engineering](https://www.anthropic.com/engineering/effective-context-engineering-for-ai-agents).

from line 78

7878<CodeGroup>

7979 ```bash cURL

8080 curl https://api.anthropic.com/v1/messages \

81 --header "x-api-key: $ANTHROPIC_API_KEY" \

82 --header "anthropic-version: 2023-06-01" \

83 --header "content-type: application/json" \

84 --data '{

81 -H "x-api-key: $ANTHROPIC_API_KEY" \

82 -H "anthropic-version: 2023-06-01" \

83 -H "content-type: application/json" \

84 -d '{

8585 "model": "claude-opus-5",

8686 "max_tokens": 2048,

8787 "messages": [

api/admin Changed · +11 / -11 lines

This page is larger than the 256 KiB this site keeps, so one side of the diff below stops where the stored text does.

from line 7312

73127312```http

73137313curl https://api.anthropic.com/v1/organizations/analytics/summaries \

73147314 -H 'anthropic-version: 2023-06-01' \

7315 -H "Authorization: Bearer $ANTHROPIC_OAUTH_TOKEN"

7315 -H "X-Api-Key: $ANTHROPIC_ADMIN_API_KEY"

73167316```

73177317

73187318#### Response

from line 7813

78137813```http

78147814curl https://api.anthropic.com/v1/organizations/analytics/usage_report \

78157815 -H 'anthropic-version: 2023-06-01' \

7816 -H "Authorization: Bearer $ANTHROPIC_OAUTH_TOKEN"

7816 -H "X-Api-Key: $ANTHROPIC_ADMIN_API_KEY"

78177817```

78187818

78197819#### Response

from line 8128

81288128```http

81298129curl https://api.anthropic.com/v1/organizations/analytics/user_usage_report \

81308130 -H 'anthropic-version: 2023-06-01' \

8131 -H "Authorization: Bearer $ANTHROPIC_OAUTH_TOKEN"

8131 -H "X-Api-Key: $ANTHROPIC_ADMIN_API_KEY"

81328132```

81338133

81348134#### Response

from line 8658

86588658```http

86598659curl https://api.anthropic.com/v1/organizations/analytics/cost_report \

86608660 -H 'anthropic-version: 2023-06-01' \

8661 -H "Authorization: Bearer $ANTHROPIC_OAUTH_TOKEN"

8661 -H "X-Api-Key: $ANTHROPIC_ADMIN_API_KEY"

86628662```

86638663

86648664#### Response

from line 8970

89708970```http

89718971curl https://api.anthropic.com/v1/organizations/analytics/user_cost_report \

89728972 -H 'anthropic-version: 2023-06-01' \

8973 -H "Authorization: Bearer $ANTHROPIC_OAUTH_TOKEN"

8973 -H "X-Api-Key: $ANTHROPIC_ADMIN_API_KEY"

89748974```

89758975

89768976#### Response

from line 9124

91249124

91259125 - `next_page: string or null`

91269126

9127 Opaque cursor for the next page, or null when `has_more

9127 Opaque cursor for the next page, or null when `has_more` is false. Pass it as the `page` parameter,

api/admin/analytics Changed · +11 / -11 lines

from line 162

162162```http

163163curl https://api.anthropic.com/v1/organizations/analytics/summaries \

164164 -H 'anthropic-version: 2023-06-01' \

165 -H "Authorization: Bearer $ANTHROPIC_OAUTH_TOKEN"

165 -H "X-Api-Key: $ANTHROPIC_ADMIN_API_KEY"

166166```

167167

168168#### Response

from line 663

663663```http

664664curl https://api.anthropic.com/v1/organizations/analytics/usage_report \

665665 -H 'anthropic-version: 2023-06-01' \

666 -H "Authorization: Bearer $ANTHROPIC_OAUTH_TOKEN"

666 -H "X-Api-Key: $ANTHROPIC_ADMIN_API_KEY"

667667```

668668

669669#### Response

from line 978

978978```http

979979curl https://api.anthropic.com/v1/organizations/analytics/user_usage_report \

980980 -H 'anthropic-version: 2023-06-01' \

981 -H "Authorization: Bearer $ANTHROPIC_OAUTH_TOKEN"

981 -H "X-Api-Key: $ANTHROPIC_ADMIN_API_KEY"

982982```

983983

984984#### Response

from line 1508

15081508```http

15091509curl https://api.anthropic.com/v1/organizations/analytics/cost_report \

15101510 -H 'anthropic-version: 2023-06-01' \

1511 -H "Authorization: Bearer $ANTHROPIC_OAUTH_TOKEN"

1511 -H "X-Api-Key: $ANTHROPIC_ADMIN_API_KEY"

15121512```

15131513

15141514#### Response

from line 1820

18201820```http

18211821curl https://api.anthropic.com/v1/organizations/analytics/user_cost_report \

18221822 -H 'anthropic-version: 2023-06-01' \

1823 -H "Authorization: Bearer $ANTHROPIC_OAUTH_TOKEN"

1823 -H "X-Api-Key: $ANTHROPIC_ADMIN_API_KEY"

18241824```

18251825

18261826#### Response

from line 2501

25012501```http

25022502curl https://api.anthropic.com/v1/organizations/analytics/users \

25032503 -H 'anthropic-version: 2023-06-01' \

2504 -H "Authorization: Bearer $ANTHROPIC_OAUTH_TOKEN"

2504 -H "X-Api-Key: $ANTHROPIC_ADMIN_API_KEY"

25052505```

25062506

25072507#### Response

from line 3136

31363136```http

31373137curl https://api.anthropic.com/v1/organizations/analytics/skills \

31383138 -H 'anthropic-version: 2023-06-01' \

3139 -H "Authorization: Bearer $ANTHROPIC_OAUTH_TOKEN"

3139 -H "X-Api-Key: $ANTHROPIC_ADMIN_API_KEY"

31403140```

31413141

31423142#### Response

from line 3481

34813481```http

34823482curl https://api.anthropic.com/v1/organizations/analytics/connectors \

34833483 -H 'anthropic-version: 2023-06-01' \

3484 -H "Authorization: Bearer $ANTHROPIC_OAUTH_TOKEN"

3484 -H "X-Api-Key: $ANTHROPIC_ADMIN_API_KEY"

34853485```

34863486

34873487#### Response

from line 3776

37763776```http

37773777curl https://api.anthropic.com/v1/organizations/analytics/apps/chat/projects \

37783778 -H 'anthropic-version: 2023-06-01' \

3779 -H "Authorization: Bearer $ANTHROPIC_OAUTH_TOKEN"

3779 -H "X-Api-Key: $ANTHROPIC_ADMIN_API_KEY"

37803780```

37813781

37823782#### Response

from line 4015

40154015```http

40164016curl https://api.anthropic.com/v1/organizations/analytics/plugins \

40174017 -H 'anthropic-version: 2023-06-01' \

4018 -H "Authorization: Bearer $ANTHROPIC_OAUTH_TOKEN"

4018 -H "X-Api-Key: $ANTHROPIC_ADMIN_API_KEY"

40194019```

40204020

40214021#### Response

from line 4208

42084208```http

42094209curl https://api.anthropic.com/v1/organizations/analytics/artifacts \

42104210 -H 'anthropic-version: 2023-06-01' \

4211 -H "Authorization: Bearer $ANTHROPIC_OAUTH_TOKEN"

4211 -H "X-Api-Key: $ANTHROPIC_ADMIN_API_KEY"

42124212```

42134213

42144214#### Response

api/beta Changed · +19823 / -10786 lines

### Beta Managed Agents Bash Tool Config ### Beta Managed Agents Bash Tool Config Params ### Beta Managed Agents Edit Tool Config ### Beta Managed Agents Edit Tool Config Params ### Beta Managed Agents Glob Tool Config ### Beta Managed Agents Glob Tool Config Params ### Beta Managed Agents Grep Tool Config ### Beta Managed Agents Grep Tool Config Params ### Beta Managed Agents Read Tool Config ### Beta Managed Agents Read Tool Config Params ### Beta Managed Agents User Location ### Beta Managed Agents Web Fetch Tool Config ### Beta Managed Agents Web Fetch Tool Config Params ### Beta Managed Agents Web Search Tool Config ### Beta Managed Agents Web Search Tool Config Params ### Beta Managed Agents Write Tool Config ### Beta Managed Agents Write Tool Config Params

This page is larger than the 256 KiB this site keeps, so one side of the diff below stops where the stored text does.

from line 5363

53635363

53645364 - `version: string`

53655365

5366 Skill version or 'latest' for most recent version

5366 The resolved version: a skill version ID for custom skills.

53675367

53685368 - `content: array of BetaContentBlock`

53695369

from line 8702

87028702

87038703- `output_format: optional BetaJSONOutputFormat or null`

87048704

8705 Deprecated: Use `output_config.format` instead. See [structured outputs](https://platform.claude.com/

8705 Deprecated: Use `output_config.format` instead. See [structured outputs](https://platform.c

api/beta/agents Changed · +4525 / -701 lines

This page is larger than the 256 KiB this site keeps, so one side of the diff below stops where the stored text does.

The two sides of this change are more than 400 edits apart, too far apart to line up, so this is the differ's own diff of it and the words inside a line are not marked.

from line 375

375375

376376 Per-tool configuration overrides.

377377

378 - `name: "bash" or "edit" or "read" or 5 more`

379

380 Built-in agent tool identifier.

381

382 - `"bash"`

383

384 - `"edit"`

385

386 - `"read"`

387

388 - `"write"`

389

390 - `"glob"`

391

392 - `"grep"`

393

394 - `"web_fetch"`

395

396 - `"web_search"`

378 - `BetaManagedAgentsBashToolConfigParams object { name, enabled, permission_policy, type }`

379

380 Configuration override for the bash tool.

381

382 - `name: "bash"`

383

384 Must be "bash".

385

386 - `"bash"`

387

388 - `enabled: optional boolean or null`

389

390 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

391

392 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

393

394 Permission policy for tool execution.

395

396 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

397

398 Tool calls are automatically approved without user confirmation.

399

400 - `type: "always_allow"`

401

402 - `"always_allow"`

403

404 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

405

406 Tool calls require user confirmation before execution.

407

408 - `type: "always_ask"`

409

410 - `"always_ask"`

411

412 - `type: optional "bash"`

413

414 - `"bash"`

415

416 - `BetaManagedAgentsEditToolConfigParams object { name, enabled, permission_policy, type }`

417

418 Configuration override for the edit tool.

419

420 - `name: "edit"`

421

422 Must be "edit".

423

424 - `"edit"`

425

426 - `enabled: optional boolean or null`

427

428 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

429

430 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

431

432 Permission policy for tool execution.

433

434 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

435

436 Tool calls are automatically approved without user confirmation.

437

438 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

439

440 Tool calls require user confirmation before execution.

441

442 - `type: optional "edit"`

443

444 - `"edit"`

445

446 - `BetaManagedAgentsReadToolConfigParams object { name, enabled, permission_policy, type }`

447

448 Configuration override for the read tool.

449

450 - `name: "read"`

451

452 Must be "read".

453

454 - `"read"`

455

456 - `enabled: optional boolean or null`

457

458 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

459

460 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

461

462 Permission policy for tool execution.

463

464 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

465

466 Tool calls are automatically approved without user confirmation.

467

468 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

469

470 Tool calls require user confirmation before execution.

471

472 - `type: optional "read"`

473

474 - `"read"`

475

476 - `BetaManagedAgentsWriteToolConfigParams object { name, enabled, permission_policy, type }`

477

478 Configuration override for the write tool.

479

480 - `name: "write"`

481

482 Must be "write".

483

484 - `"write"`

485

486 - `enabled: optional boolean or null`

487

488 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

489

490 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

491

492 Permission policy for tool execution.

493

494 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

495

496 Tool calls are automatically approved without user confirmation.

497

498 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

499

500 Tool calls require user confirmation before execution.

501

502 - `type: optional "write"`

503

504 - `"write"`

505

506 - `BetaManagedAgentsGlobToolConfigParams object { name, enabled, permission_policy, type }`

507

508 Configuration override for the glob tool.

509

510 - `name: "glob"`

511

512 Must be "glob".

513

514 - `"glob"`

515

516 - `enabled: optional boolean or null`

517

518 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

519

520 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

521

522 Permission policy for tool execution.

523

524 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

525

526 Tool calls are automatically approved without user confirmation.

527

528 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

529

530 Tool calls require user confirmation before execution.

531

532 - `type: optional "glob"`

533

534 - `"glob"`

535

536 - `BetaManagedAgentsGrepToolConfigParams object { name, enabled, permission_policy, type }`

537

538 Configuration override for the grep tool.

539

540 - `name: "grep"`

541

542 Must be "grep".

543

544 - `"grep"`

545

546 - `enabled: optional boolean or null`

547

548 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

549

550 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

551

552 Permission policy for tool execution.

553

554 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

555

556 Tool calls are automatically approved without user confirmation.

557

558 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

559

560 Tool calls require user confirmation before execution.

561

562 - `type: optional "grep"`

563

564 - `"grep"`

565

566 - `BetaManagedAgentsWebFetchToolConfigParams object { name, allowed_domains, blocked_domains, 4 more }`

567

568 Configuration override for the web_fetch tool.

569

570 - `name: "web_fetch"`

571

572 Must be "web_fetch".

573

574 - `"web_fetch"`

575

576 - `allowed_domains: optional array of string`

577

578 Only fetch URLs whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "docs.example.com" (no scheme, port, or path). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with blocked_domains.

579

580 - `blocked_domains: optional array of string`

581

582 Never fetch URLs whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "ads.example.com" (no scheme, port, or path). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with allowed_domains.

583

584 - `enabled: optional boolean or null`

585

586 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

587

588 - `max_content_tokens: optional number or null`

589

590 Maximum number of tokens of fetched text content to include in context per call. Does not apply to binary content such as PDFs.

591

592 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

593

594 Permission policy for tool execution.

595

596 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

597

598 Tool calls are automatically approved without user confirmation.

599

600 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

601

602 Tool calls require user confirmation before execution.

603

604 - `type: optional "web_fetch"`

605

606 - `"web_fetch"`

607

608 - `BetaManagedAgentsWebSearchToolConfigParams object { name, allowed_domains, blocked_domains, 4 more }`

609

610 Configuration override for the web_search tool.

611

612 - `name: "web_search"`

613

614 Must be "web_search".

615

616 - `"web_search"`

617

618 - `allowed_domains: optional array of string`

619

620 Only return search results whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "docs.example.com" (no scheme or port; an optional path suffix is accepted). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with blocked_domains.

621

622 - `blocked_domains: optional array of string`

623

624 Never return search results whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "ads.example.com" (no scheme or port; an optional path suffix is accepted). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with allowed_domains.

625

626 - `enabled: optional boolean or null`

627

628 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

629

630 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

631

632 Permission policy for tool execution.

633

634 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

635

636 Tool calls are automatically approved without user confirmation.

637

638 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

639

640 Tool calls require user confirmation before execution.

641

642 - `type: optional "web_search"`

643

644 - `"web_search"`

645

646 - `user_location: optional BetaManagedAgentsUserLocation or null`

647

648 Approximate user location for search result localization.

649

650 - `type: "approximate"`

651

652 Location precision. Only "approximate" is supported.

653

654 - `"approximate"`

655

656 - `city: optional string or null`

657

658 City name.

659

660 - `country: optional string or null`

661

662 Two-letter ISO 3166-1 country code, uppercase.

663

664 - `region: optional string or null`

665

666 Region or state name.

667

668 - `timezone: optional string or null`

669

670 IANA timezone identifier, e.g. "America/Los_Angeles".

671

672 - `default_config: optional BetaManagedAgentsAgentToolsetDefaultConfigParams or null`

673

674 Default configuration for all tools in a toolset.

397675

398676 - `enabled: optional boolean or null`

399677

400 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

678 Whether tools are enabled and available to Claude by default. Defaults to true if not specified.

401679

402680 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

403681

from line 685

407685

408686 Tool calls are automatically approved without user confirmation.

409687

410 - `type: "always_allow"`

411

412 - `"always_allow"`

413

414688 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

415689

416690 Tool calls require user confirmation before execution.

417691

418 - `type: "always_ask"`

419

420 - `"always_ask"`

421

422 - `default_config: optional BetaManagedAgentsAgentToolsetDefaultConfigParams or null`

423

424 Default configuration for all tools in a toolset.

692 - `BetaManagedAgentsMCPToolsetParams object { mcp_server_name, type, configs, default_config }`

693

694 Configuration for tools from an MCP server defined in `mcp_servers`.

695

696 - `mcp_server_name: string`

697

698 Name of the MCP server. Must match a server name from the mcp_servers array. 1-255 characters.

699

700 - `type: "mcp_toolset"`

701

702 - `"mcp_toolset"`

703

704 - `configs: optional array of BetaManagedAgentsMCPToolConfigParams`

705

706 Per-tool configuration overrides.

707

708 - `name: string`

709

710 Name of the MCP tool to configure. 1-128 characters.

425711

426712 - `enabled: optional boolean or null`

427713

428 Whether tools are enabled and available to Claude by default. Defaults to true if not specified.

714 Whether this tool is enabled. Overrides the `default_config` setting.

429715

430716 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

431717

from line 725

439725

440726 Tool calls require user confirmation before execution.

441727

442 - `BetaManagedAgentsMCPToolsetParams object { mcp_server_name, type, configs, default_config }`

443

444 Configuration for tools from an MCP server defined in `mcp_servers`.

445

446 - `mcp_server_name: string`

447

448 Name of the MCP server. Must match a server name from the mcp_servers array. 1-255 characters.

449

450 - `type: "mcp_toolset"`

451

452 - `"mcp_toolset"`

453

454 - `configs: optional array of BetaManagedAgentsMCPToolConfigParams`

455

456 Per-tool configuration overrides.

457

458 - `name: string`

459

460 Name of the MCP tool to configure. 1-128 characters.

728 - `default_config: optional BetaManagedAgentsMCPToolsetDefaultConfigParams or null`

729

730 Default configuration for all tools from an MCP server.

461731

462732 - `enabled: optional boolean or null`

463733

464 Whether this tool is enabled. Overrides the `default_config` setting.

734 Whether tools are enabled by default. Defaults to true if not specified.

465735

466736 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

467737

from line 745

475745

476746 Tool calls require user confirmation before execution.

477747

478 - `default_config: optional BetaManagedAgentsMCPToolsetDefaultConfigParams or null`

479

480 Default configuration for all tools from an MCP server.

481

482 - `enabled: optional boolean or null`

483

484 Whether tools are enabled by default. Defaults to true if not specified.

485

486 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

487

488 Permission policy for tool execution.

489

490 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

491

492 Tool calls are automatically approved without user confirmation.

493

494 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

495

496 Tool calls require user confirmation before execution.

497

498748 - `BetaManagedAgentsCustomToolParams object { description, input_schema, name, type }`

499749

500750 A custom tool that is executed by the API client rather than the agent. When the agent calls this tool, an `agent.custom_tool_use` event is emitted and the session goes idle, waiting for the client to provide the result via a `user.custom_tool_result` event.

from line 1001

7511001

7521002 - `configs: array of BetaManagedAgentsAgentToolConfig`

7531003

754 - `enabled: boolean`

755

756 - `name: "bash" or "edit" or "read" or 5 more`

757

758 Built-in agent tool identifier.

759

760 - `"bash"`

761

762 - `"edit"`

763

764 - `"read"`

765

766 - `"write"`

767

768 - `"glob"`

769

770 - `"grep"`

771

772 - `"web_fetch"`

773

774 - `"web_search"`

775

776 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

777

778 Permission policy for tool execution.

779

780 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

781

782 Tool calls are automatically approved without user confirmation.

783

784 - `type: "always_allow"`

785

786 - `"always_allow"`

787

788 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

789

790 Tool calls require user confirmation before execution.

791

792 - `type: "always_ask"`

793

794 - `"always_ask"`

1004 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

1005

1006 Configuration for the bash tool.

1007

1008 - `enabled: boolean`

1009

1010 - `name: "bash"`

1011

1012 - `"bash"`

1013

1014 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1015

1016 Permission policy for tool execution.

1017

1018 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1019

1020 Tool calls are automatically approved without user confirmation.

1021

1022 - `type: "always_allow"`

1023

1024 - `"always_allow"`

1025

1026 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1027

1028 Tool calls require user confirmation before execution.

1029

1030 - `type: "always_ask"`

1031

1032 - `"always_ask"`

1033

1034 - `type: "bash"`

1035

1036 - `"bash"`

1037

1038 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

1039

1040 Configuration for the edit tool.

1041

1042 - `enabled: boolean`

1043

1044 - `name: "edit"`

1045

1046 - `"edit"`

1047

1048 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1049

1050 Permission policy for tool execution.

1051

1052 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1053

1054 Tool calls are automatically approved without user confirmation.

1055

1056 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1057

1058 Tool calls require user confirmation before execution.

1059

1060 - `type: "edit"`

1061

1062 - `"edit"`

1063

1064 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

1065

1066 Configuration for the read tool.

1067

1068 - `enabled: boolean`

1069

1070 - `name: "read"`

1071

1072 - `"read"`

1073

1074 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1075

1076 Permission policy for tool execution.

1077

1078 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1079

1080 Tool calls are automatically approved without user confirmation.

1081

1082 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1083

1084 Tool calls require user confirmation before execution.

1085

1086 - `type: "read"`

1087

1088 - `"read"`

1089

1090 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

1091

1092 Configuration for the write tool.

1093

1094 - `enabled: boolean`

1095

1096 - `name: "write"`

1097

1098 - `"write"`

1099

1100 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1101

1102 Permission policy for tool execution.

1103

1104 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1105

1106 Tool calls are automatically approved without user confirmation.

1107

1108 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1109

1110 Tool calls require user confirmation before execution.

1111

1112 - `type: "write"`

1113

1114 - `"write"`

1115

1116 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

1117

1118 Configuration for the glob tool.

1119

1120 - `enabled: boolean`

1121

1122 - `name: "glob"`

1123

1124 - `"glob"`

1125

1126 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1127

1128 Permission policy for tool execution.

1129

1130 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1131

1132 Tool calls are automatically approved without user confirmation.

1133

1134 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1135

1136 Tool calls require user confirmation before execution.

1137

1138 - `type: "glob"`

1139

1140 - `"glob"`

1141

1142 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

1143

1144 Configuration for the grep tool.

1145

1146 - `enabled: boolean`

1147

1148 - `name: "grep"`

1149

1150 - `"grep"`

1151

1152 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1153

1154 Permission policy for tool execution.

1155

1156 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1157

1158 Tool calls are automatically approved without user confirmation.

1159

1160 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1161

1162 Tool calls require user confirmation before execution.

1163

1164 - `type: "grep"`

1165

1166 - `"grep"`

1167

1168 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

1169

1170 Configuration for the web_fetch tool.

1171

1172 - `enabled: boolean`

1173

1174 - `name: "web_fetch"`

1175

1176 - `"web_fetch"`

1177

1178 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1179

1180 Permission policy for tool execution.

1181

1182 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1183

1184 Tool calls are automatically approved without user confirmation.

1185

1186 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1187

1188 Tool calls require user confirmation before execution.

1189

1190 - `type: "web_fetch"`

1191

1192 - `"web_fetch"`

1193

1194 - `allowed_domains: optional array of string`

1195

1196 - `blocked_domains: optional array of string`

1197

1198 - `max_content_tokens: optional number or null`

1199

1200 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

1201

1202 Configuration for the web_search tool.

1203

1204 - `enabled: boolean`

1205

1206 - `name: "web_search"`

1207

1208 - `"web_search"`

1209

1210 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1211

1212 Permission policy for tool execution.

1213

1214 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1215

1216 Tool calls are automatically approved without user confirmation.

1217

1218 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1219

1220 Tool calls require user confirmation before execution.

1221

1222 - `type: "web_search"`

1223

1224 - `"web_search"`

1225

1226 - `allowed_domains: optional array of string`

1227

1228 - `blocked_domains: optional array of string`

1229

1230 - `user_location: optional BetaManagedAgentsUserLocation or null`

1231

1232 Approximate user location for search result localization.

1233

1234 - `type: "approximate"`

1235

1236 Location precision. Only "approximate" is supported.

1237

1238 - `"approximate"`

1239

1240 - `city: optional string or null`

1241

1242 City name.

1243

1244 - `country: optional string or null`

1245

1246 Two-letter ISO 3166-1 country code, uppercase.

1247

1248 - `region: optional string or null`

1249

1250 Region or state name.

1251

1252 - `timezone: optional string or null`

1253

1254 IANA timezone identifier, e.g. "America/Los_Angeles".

7951255

7961256 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

7971257

from line 1437

9771437 "name": "bash",

9781438 "permission_policy": {

9791439 "type": "always_allow"

980 }

1440 },

1441 "type": "bash"

9811442 }

9821443 ],

9831444 "default_config": {

from line 1790

13291790

13301791 - `configs: array of BetaManagedAgentsAgentToolConfig`

13311792

1332 - `enabled: boolean`

1333

1334 - `name: "bash" or "edit" or "read" or 5 more`

1335

1336 Built-in agent tool identifier.

1337

1338 - `"bash"`

1339

1340 - `"edit"`

1341

1342 - `"read"`

1343

1344 - `"write"`

1345

1346 - `"glob"`

1347

1348 - `"grep"`

1349

1350 - `"web_fetch"`

1351

1352 - `"web_search"`

1353

1354 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1355

1356 Permission policy for tool execution.

1357

1358 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1359

1360 Tool calls are automatically approved without user confirmation.

1361

1362 - `type: "always_allow"`

1363

1364 - `"always_allow"`

1365

1366 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1367

1368 Tool calls require user confirmation before execution.

1369

1370 - `type: "always_ask"`

1371

1372 - `"always_ask"`

1793 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

1794

1795 Configuration for the bash tool.

1796

1797 - `enabled: boolean`

1798

1799 - `name: "bash"`

1800

1801 - `"bash"`

1802

1803 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1804

1805 Permission policy for tool execution.

1806

1807 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1808

1809 Tool calls are automatically approved without user confirmation.

1810

1811 - `type: "always_allow"`

1812

1813 - `"always_allow"`

1814

1815 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1816

1817 Tool calls require user confirmation before execution.

1818

1819 - `type: "always_ask"`

1820

1821 - `"always_ask"`

1822

1823 - `type: "bash"`

1824

1825 - `"bash"`

1826

1827 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

1828

1829 Configuration for the edit tool.

1830

1831 - `enabled: boolean`

1832

1833 - `name: "edit"`

1834

1835 - `"edit"`

1836

1837 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1838

1839 Permission policy for tool execution.

1840

1841 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1842

1843 Tool calls are automatically approved without user confirmation.

1844

1845 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1846

1847 Tool calls require user confirmation before execution.

1848

1849 - `type: "edit"`

1850

1851 - `"edit"`

1852

1853 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

1854

1855 Configuration for the read tool.

1856

1857 - `enabled: boolean`

1858

1859 - `name: "read"`

1860

1861 - `"read"`

1862

1863 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1864

1865 Permission policy for tool execution.

1866

1867 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1868

1869 Tool calls are automatically approved without user confirmation.

1870

1871 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1872

1873 Tool calls require user confirmation before execution.

1874

1875 - `type: "read"`

1876

1877 - `"read"`

1878

1879 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

1880

1881 Configuration for the write tool.

1882

1883 - `enabled: boolean`

1884

1885 - `name: "write"`

1886

1887 - `"write"`

1888

1889 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1890

1891 Permission policy for tool execution.

1892

1893 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1894

1895 Tool calls are automatically approved without user confirmation.

1896

1897 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1898

1899 Tool calls require user confirmation before execution.

1900

1901 - `type: "write"`

1902

1903 - `"write"`

1904

1905 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

1906

1907 Configuration for the glob tool.

1908

1909 - `enabled: boolean`

1910

1911 - `name: "glob"`

1912

1913 - `"glob"`

1914

1915 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1916

1917 Permission policy for tool execution.

1918

1919 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1920

1921 Tool calls are automatically approved without user confirmation.

1922

1923 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1924

1925 Tool calls require user confirmation before execution.

1926

1927 - `type: "glob"`

1928

1929 - `"glob"`

1930

1931 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

1932

1933 Configuration for the grep tool.

1934

1935 - `enabled: boolean`

1936

1937 - `name: "grep"`

1938

1939 - `"grep"`

1940

1941 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1942

1943 Permission policy for tool execution.

1944

1945 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1946

1947 Tool calls are automatically approved without user confirmation.

1948

1949 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1950

1951 Tool calls require user confirmation before execution.

1952

1953 - `type: "grep"`

1954

1955 - `"grep"`

1956

1957 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

1958

1959 Configuration for the web_fetch tool.

1960

1961 - `enabled: boolean`

1962

1963 - `name: "web_fetch"`

1964

1965 - `"web_fetch"`

1966

1967 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1968

1969 Permission policy for tool execution.

1970

1971 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1972

1973 Tool calls are automatically approved without user confirmation.

1974

1975 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1976

1977 Tool calls require user confirmation before execution.

1978

1979 - `type: "web_fetch"`

1980

1981 - `"web_fetch"`

1982

1983 - `allowed_domains: optional array of string`

1984

1985 - `blocked_domains: optional array of string`

1986

1987 - `max_content_tokens: optional number or null`

1988

1989 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

1990

1991 Configuration for the web_search tool.

1992

1993 - `enabled: boolean`

1994

1995 - `name: "web_search"`

1996

1997 - `"web_search"`

1998

1999 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

2000

2001 Permission policy for tool execution.

2002

2003 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

2004

2005 Tool calls are automatically approved without user confirmation.

2006

2007 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

2008

2009 Tool calls require user confirmation before execution.

2010

2011 - `type: "web_search"`

2012

2013 - `"web_search"`

2014

2015 - `allowed_domains: optional array of string`

2016

2017 - `blocked_domains: optional array of string`

2018

2019 - `user_location: optional BetaManagedAgentsUserLocation or null`

2020

2021 Approximate user location for search result localization.

2022

2023 - `type: "approximate"`

2024

2025 Location precision. Only "approximate" is supported.

2026

2027 - `"approximate"`

2028

2029 - `city: optional string or null`

2030

2031 City name.

2032

2033 - `country: optional string or null`

2034

2035 Two-letter ISO 3166-1 country code, uppercase.

2036

2037 - `region: optional string or null`

2038

2039 Region or state name.

2040

2041 - `timezone: optional string or null`

2042

2043 IANA timezone identifier, e.g. "America/Los_Angeles".

13732044

13742045 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

13752046

from line 2217

15462217 "name": "bash",

15472218 "permission_policy": {

15482219 "type": "always_allow"

1549 }

2220 },

2221 "type": "bash"

15502222 }

15512223 ],

15522224 "default_config": {

from line 2561

18892561

18902562 - `configs: array of BetaManagedAgentsAgentToolConfig`

18912563

1892 - `enabled: boolean`

1893

1894 - `name: "bash" or "edit" or "read" or 5 more`

1895

1896 Built-in agent tool identifier.

1897

1898 - `"bash"`

1899

1900 - `"edit"`

1901

1902 - `"read"`

1903

1904 - `"write"`

1905

1906 - `"glob"`

1907

1908 - `"grep"`

1909

1910 - `"web_fetch"`

1911

1912 - `"web_search"`

1913

1914 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1915

1916 Permission policy for tool execution.

1917

1918 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1919

1920 Tool calls are automatically approved without user confirmation.

1921

1922 - `type: "always_allow"`

1923

1924 - `"always_allow"`

1925

1926 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1927

1928 Tool calls require user confirmation before execution.

1929

1930 - `type: "always_ask"`

1931

1932 - `"always_ask"`

2564 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

2565

2566 Configuration for the bash tool.

2567

2568 - `enabled: boolean`

2569

2570 - `name: "bash"`

2571

2572 - `"bash"`

2573

2574 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

2575

2576 Permission policy for tool execution.

2577

2578 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

2579

2580 Tool calls are automatically approved without user confirmation.

2581

2582 - `type: "always_allow"`

2583

2584 - `"always_allow"`

2585

2586 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

2587

2588 Tool calls require user confirmation before execution.

2589

2590 - `type: "always_ask"`

2591

2592 - `"always_ask"`

2593

2594 - `type: "bash"`

2595

2596 - `"bash"`

2597

2598 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

2599

2600 Configuration for the edit tool.

2601

2602 - `enabled: boolean`

2603

2604 - `name: "edit"`

2605

2606 - `"edit"`

2607

2608 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

2609

2610 Permission policy for tool execution.

2611

2612 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

2613

2614 Tool calls are automatically approved without user confirmation.

2615

2616 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

2617

2618 Tool calls require user confirmation before execution.

2619

2620 - `type: "edit"`

2621

2622 - `"edit"`

2623

2624 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

2625

2626 Configuration for the read tool.

2627

2628 - `enabled: boolean`

2629

2630 - `name: "read"`

2631

2632 - `"read"`

2633

2634 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

2635

2636 Permission policy for tool execution.

2637

2638 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

2639

2640 Tool calls are automatically approved without user confirmation.

2641

2642 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

2643

2644 Tool calls require user confirmation before execution.

2645

2646 - `type: "read"`

2647

2648 - `"read"`

2649

2650 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

2651

2652 Configuration for the write tool.

2653

2654 - `enabled: boolean`

2655

2656 - `name: "write"`

2657

2658 - `"write"`

2659

2660 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

2661

2662 Permission policy for tool execution.

2663

2664 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

2665

2666 Tool calls are automatically approved without user confirmation.

2667

2668 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

2669

2670 Tool calls require user confirmation before execution.

2671

2672 - `type: "write"`

2673

2674 - `"write"`

2675

2676 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

2677

2678 Configuration for the glob tool.

2679

2680 - `enabled: boolean`

2681

2682 - `name: "glob"`

2683

2684 - `"glob"`

2685

2686 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

2687

2688 Permission policy for tool execution.

2689

2690 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

2691

2692 Tool calls are automatically approved without user confirmation.

2693

2694 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

2695

2696 Tool calls require user confirmation before execution.

2697

2698 - `type: "glob"`

2699

2700 - `"glob"`

2701

2702 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

2703

2704 Configuration for the grep tool.

2705

2706 - `enabled: boolean`

2707

2708 - `name: "grep"`

2709

2710 - `"grep"`

2711

2712 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

2713

2714 Permission policy for tool execution.

2715

2716 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

2717

2718 Tool calls are automatically approved without user confirmation.

2719

2720 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

2721

2722 Tool calls require user confirmation before execution.

2723

2724 - `type: "grep"`

2725

2726 - `"grep"`

2727

2728 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

2729

2730 Configuration for the web_fetch tool.

2731

2732 - `enabled: boolean`

2733

2734 - `name: "web_fetch"`

2735

2736 - `"web_fetch"`

2737

2738 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

2739

2740 Permission policy for tool execution.

2741

2742 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

2743

2744 Tool calls are automatically approved without user confirmation.

2745

2746 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

2747

2748 Tool calls require user confirmation before execution.

2749

2750 - `type: "web_fetch"`

2751

2752 - `"web_fetch"`

2753

2754 - `allowed_domains: optional array of string`

2755

2756 - `blocked_domains: optional array of string`

2757

2758 - `max_content_tokens: optional number or null`

2759

2760 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

2761

2762 Configuration for the web_search tool.

2763

2764 - `enabled: boolean`

2765

2766 - `name: "web_search"`

2767

2768 - `"web_search"`

2769

2770 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

2771

2772 Permission policy for tool execution.

2773

2774 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

2775

2776 Tool calls are automatically approved without user confirmation.

2777

2778 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

2779

2780 Tool calls require user confirmation before execution.

2781

2782 - `type: "web_search"`

2783

2784 - `"web_search"`

2785

2786 - `allowed_domains: optional array of string`

2787

2788 - `blocked_domains: optional array of string`

2789

2790 - `user_location: optional BetaManagedAgentsUserLocation or null`

2791

2792 Approximate user location for search result localization.

2793

2794 - `type: "approximate"`

2795

2796 Location precision. Only "approximate" is supported.

2797

2798 - `"approximate"`

2799

2800 - `city: optional string or null`

2801

2802 City name.

2803

2804 - `country: optional string or null`

2805

2806 Two-letter ISO 3166-1 country code, uppercase.

2807

2808 - `region: optional string or null`

2809

2810 Region or state name.

2811

2812 - `timezone: optional string or null`

2813

2814 IANA timezone identifier, e.g. "America/Los_Angeles".

19332815

19342816 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

19352817

from line 2982

21002982 "name": "bash",

21012983 "permission_policy": {

21022984 "type": "always_allow"

2103 }

2985 },

2986 "type": "bash"

21042987 }

21052988 ],

21062989 "default_config": {

from line 3375

24923375

24933376 Per-tool configuration overrides.

24943377

2495 - `name: "bash" or "edit" or "read" or 5 more`

2496

2497 Built-in agent tool identifier.

2498

2499 - `"bash"`

2500

2501 - `"edit"`

2502

2503 - `"read"`

2504

2505 - `"write"`

2506

2507 - `"glob"`

2508

2509 - `"grep"`

2510

2511 - `"web_fetch"`

2512

2513 - `"web_search"`

3378 - `BetaManagedAgentsBashToolConfigParams object { name, enabled, permission_policy, type }`

3379

3380 Configuration override for the bash tool.

3381

3382 - `name: "bash"`

3383

3384 Must be "bash".

3385

3386 - `"bash"`

3387

3388 - `enabled: optional boolean or null`

3389

3390 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

3391

3392 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

3393

3394 Permission policy for tool execution.

3395

3396 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

3397

3398 Tool calls are automatically approved without user confirmation.

3399

3400 - `type: "always_allow"`

3401

3402 - `"always_allow"`

3403

3404 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

3405

3406 Tool calls require user confirmation before execution.

3407

3408 - `type: "always_ask"`

3409

3410 - `"always_ask"`

3411

3412 - `type: optional "bash"`

3413

3414 - `"bash"`

3415

3416 - `BetaManagedAgentsEditToolConfigParams object { name, enabled, permission_policy, type }`

3417

3418 Configuration override for the edit tool.

3419

3420 - `name: "edit"`

3421

3422 Must be "edit".

3423

3424 - `"edit"`

3425

3426 - `enabled: optional boolean or null`

3427

3428 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

3429

3430 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

3431

3432 Permission policy for tool execution.

3433

3434 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

3435

3436 Tool calls are automatically approved without user confirmation.

3437

3438 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

3439

3440 Tool calls require user confirmation before execution.

3441

3442 - `type: optional "edit"`

3443

3444 - `"edit"`

3445

3446 - `BetaManagedAgentsReadToolConfigParams object { name, enabled, permission_policy, type }`

3447

3448 Configuration override for the read tool.

3449

3450 - `name: "read"`

3451

3452 Must be "read".

3453

3454 - `"read"`

3455

3456 - `enabled: optional boolean or null`

3457

3458 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

3459

3460 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

3461

3462 Permission policy for tool execution.

3463

3464 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

3465

3466 Tool calls are automatically approved without user confirmation.

3467

3468 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

3469

3470 Tool calls require user confirmation before execution.

3471

3472 - `type: optional "read"`

3473

3474 - `"read"`

3475

3476 - `BetaManagedAgentsWriteToolConfigParams object { name, enabled, permission_policy, type }`

3477

3478 Configuration override for the write tool.

3479

3480 - `name: "write"`

3481

3482 Must be "write".

3483

3484 - `"write"`

3485

3486 - `enabled: optional boolean or null`

3487

3488 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

3489

3490 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

3491

3492 Permission policy for tool execution.

3493

3494 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

3495

3496 Tool calls are automatically approved without user confirmation.

3497

3498 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

3499

3500 Tool calls require user confirmation before execution.

3501

3502 - `type: optional "write"`

3503

3504 - `"write"`

3505

3506 - `BetaManagedAgentsGlobToolConfigParams object { name, enabled, permission_policy, type }`

3507

3508 Configuration override for the glob tool.

3509

3510 - `name: "glob"`

3511

3512 Must be "glob".

3513

3514 - `"glob"`

3515

3516 - `enabled: optional boolean or null`

3517

3518 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

3519

3520 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

3521

3522 Permission policy for tool execution.

3523

3524 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

3525

3526 Tool calls are automatically approved without user confirmation.

3527

3528 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

3529

3530 Tool calls require user confirmation before execution.

3531

3532 - `type: optional "glob"`

3533

3534 - `"glob"`

3535

3536 - `BetaManagedAgentsGrepToolConfigParams object { name, enabled, permission_policy, type }`

3537

3538 Configuration override for the grep tool.

3539

3540 - `name: "grep"`

3541

3542 Must be "grep".

3543

3544 - `"grep"`

3545

3546 - `enabled: optional boolean or null`

3547

3548 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

3549

3550 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

3551

3552 Permission policy for tool execution.

3553

3554 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

3555

3556 Tool calls are automatically approved without user confirmation.

3557

3558 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

3559

3560 Tool calls require user confirmation before execution.

3561

3562 - `type: optional "grep"`

3563

3564 - `"grep"`

3565

3566 - `BetaManagedAgentsWebFetchToolConfigParams object { name, allowed_domains, blocked_domains, 4 more }`

3567

3568 Configuration override for the web_fetch tool.

3569

3570 - `name: "web_fetch"`

3571

3572 Must be "web_fetch".

3573

3574 - `"web_fetch"`

3575

3576 - `allowed_domains: optional array of string`

3577

3578 Only fetch URLs whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "docs.example.com" (no scheme, port, or path). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with blocked_domains.

3579

3580 - `blocked_domains: optional array of string`

3581

3582 Never fetch URLs whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "ads.example.com" (no scheme, port, or path). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with allowed_domains.

3583

3584 - `enabled: optional boolean or null`

3585

3586 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

3587

3588 - `max_content_tokens: optional number or null`

3589

3590 Maximum number of tokens of fetched text content to include in context per call. Does not apply to binary content such as PDFs.

3591

3592 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

3593

3594 Permission policy for tool execution.

3595

3596 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

3597

3598 Tool calls are automatically approved without user confirmation.

3599

3600 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

3601

3602 Tool calls require user confirmation before execution.

3603

3604 - `type: optional "web_fetch"`

3605

3606 - `"web_fetch"`

3607

3608 - `BetaManagedAgentsWebSearchToolConfigParams object { name, allowed_domains, blocked_domains, 4 more }`

3609

3610 Configuration override for the web_search tool.

3611

3612 - `name: "web_search"`

3613

3614 Must be "web_search".

3615

3616 - `"web_search"`

3617

3618 - `allowed_domains: optional array of string`

3619

3620 Only return search results whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "docs.example.com" (no scheme or port; an optional path suffix is accepted). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with blocked_domains.

3621

3622 - `blocked_domains: optional array of string`

3623

3624 Never return search results whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "ads.example.com" (no scheme or port; an optional path suffix is accepted). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with allowed_domains.

3625

3626 - `enabled: optional boolean or null`

3627

3628 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

3629

3630 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

3631

3632 Permission policy for tool execution.

3633

3634 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

3635

3636 Tool calls are automatically approved without user confirmation.

3637

3638 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

3639

3640 Tool calls require user confirmation before execution.

3641

3642 - `type: optional "web_search"`

3643

3644 - `"web_search"`

3645

3646 - `user_location: optional BetaManagedAgentsUserLocation or null`

3647

3648 Approximate user location for search result localization.

3649

3650 - `type: "approximate"`

3651

3652 Location precision. Only "approximate" is supported.

3653

3654 - `"approximate"`

3655

3656 - `city: optional string or null`

3657

3658 City name.

3659

3660 - `country: optional string or null`

3661

3662 Two-letter ISO 3166-1 country code, uppercase.

3663

3664 - `region: optional string or null`

3665

3666 Region or state name.

3667

3668 - `timezone: optional string or null`

3669

3670 IANA timezone identifier, e.g. "America/Los_Angeles".

3671

3672 - `default_config: optional BetaManagedAgentsAgentToolsetDefaultConfigParams or null`

3673

3674 Default configuration for all tools in a toolset.

25143675

25153676 - `enabled: optional boolean or null`

25163677

2517 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

3678 Whether tools are enabled and available to Claude by default. Defaults to true if not specified.

25183679

25193680 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

25203681

from line 3685

25243685

25253686 Tool calls are automatically approved without user confirmation.

25263687

2527 - `type: "always_allow"`

2528

2529 - `"always_allow"`

2530

25313688 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

25323689

25333690 Tool calls require user confirmation before execution.

25343691

2535 - `type: "always_ask"`

2536

2537 - `"always_ask"`

2538

2539 - `default_config: optional BetaManagedAgentsAgentToolsetDefaultConfigParams or null`

2540

2541 Default configuration for all tools in a toolset.

3692 - `BetaManagedAgentsMCPToolsetParams object { mcp_server_name, type, configs, default_config }`

3693

3694 Configuration for tools from an MCP server defined in `mcp_servers`.

3695

3696 - `mcp_server_name: string`

3697

3698 Name of the MCP server. Must match a server name from the mcp_servers array. 1-255 characters.

3699

3700 - `type: "mcp_toolset"`

3701

3702 - `"mcp_toolset"`

3703

3704 - `configs: optional array of BetaManagedAgentsMCPToolConfigParams`

3705

3706 Per-tool configuration overrides.

3707

3708 - `name: string`

3709

3710 Name of the MCP tool to configure. 1-128 characters.

25423711

25433712 - `enabled: optional boolean or null`

25443713

2545 Whether tools are enabled and available to Claude by default. Defaults to true if not specified.

3714 Whether this tool is enabled. Overrides the `default_config` setting.

25463715

25473716 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

25483717

from line 3725

25563725

25573726 Tool calls require user confirmation before execution.

25583727

2559 - `BetaManagedAgentsMCPToolsetParams object { mcp_server_name, type, configs, default_config }`

2560

2561 Configuration for tools from an MCP server defined in `mcp_servers`.

2562

2563 - `mcp_server_name: string`

2564

2565 Name of the MCP server. Must match a server name from the mcp_servers array. 1-255 characters.

2566

2567 - `type: "mcp_toolset"`

2568

2569 - `"mcp_toolset"`

2570

2571 - `configs: optional array of BetaManagedAgentsMCPToolConfigParams`

2572

2573 Per-tool configuration overrides.

2574

2575 - `name: string`

2576

2577 Name of the MCP tool to configure. 1-128 characters.

3728 - `default_config: optional BetaManagedAgentsMCPToolsetDefaultConfigParams or null`

3729

3730 Default configuration for all tools from an MCP server.

25783731

25793732 - `enabled: optional boolean or null`

25803733

2581 Whether this tool is enabled. Overrides the `default_config` setting.

3734 Whether tools are enabled by default. Defaults to true if not specified.

25823735

25833736 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

25843737

from line 3745

25923745

25933746 Tool calls require user confirmation before execution.

25943747

2595 - `default_config: optional BetaManagedAgentsMCPToolsetDefaultConfigParams or null`

2596

2597 Default configuration for all tools from an MCP server.

2598

2599 - `enabled: optional boolean or null`

2600

2601 Whether tools are enabled by default. Defaults to true if not specified.

2602

2603 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

2604

2605 Permission policy for tool execution.

2606

2607 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

2608

2609 Tool calls are automatically approved without user confirmation.

2610

2611 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

2612

2613 Tool calls require user confirmation before execution.

2614

26153748 - `BetaManagedAgentsCustomToolParams object { description, input_schema, name, type }`

26163749

26173750 A custom tool that is executed by the API client rather than the agent. When the agent calls this tool, an `agent.custom_tool_use` event is emitted and the session goes idle, waiting for the client to provide the result via a `user.custom_tool_result` event.

from line 4005

28724005

28734006 - `configs: array of BetaManagedAgentsAgentToolConfig`

28744007

2875 - `enabled: boolean`

2876

2877 - `name: "bash" or "edit" or "read" or 5 more`

2878

2879 Built-in agent tool identifier.

2880

2881 - `"bash"`

2882

2883 - `"edit"`

2884

2885 - `"read"`

2886

2887 - `"write"`

2888

2889 - `"glob"`

2890

2891 - `"grep"`

2892

2893 - `"web_fetch"`

2894

2895 - `"web_search"`

2896

2897 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

2898

2899 Permission policy for tool execution.

2900

2901 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

2902

2903 Tool calls are automatically approved without user confirmation.

2904

2905 - `type: "always_allow"`

2906

2907 - `"always_allow"`

2908

2909 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

2910

2911 Tool calls require user confirmation before execution.

2912

2913 - `type: "always_ask"`

2914

2915 - `"always_ask"`

4008 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

4009

4010 Configuration for the bash tool.

4011

4012 - `enabled: boolean`

4013

4014 - `name: "bash"`

4015

4016 - `"bash"`

4017

4018 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

4019

4020 Permission policy for tool execution.

4021

4022 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4023

4024 Tool calls are automatically approved without user confirmation.

4025

4026 - `type: "always_allow"`

4027

4028 - `"always_allow"`

4029

4030 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4031

4032 Tool calls require user confirmation before execution.

4033

4034 - `type: "always_ask"`

4035

4036 - `"always_ask"`

4037

4038 - `type: "bash"`

4039

4040 - `"bash"`

4041

4042 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

4043

4044 Configuration for the edit tool.

4045

4046 - `enabled: boolean`

4047

4048 - `name: "edit"`

4049

4050 - `"edit"`

4051

4052 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

4053

4054 Permission policy for tool execution.

4055

4056 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4057

4058 Tool calls are automatically approved without user confirmation.

4059

4060 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4061

4062 Tool calls require user confirmation before execution.

4063

4064 - `type: "edit"`

4065

4066 - `"edit"`

4067

4068 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

4069

4070 Configuration for the read tool.

4071

4072 - `enabled: boolean`

4073

4074 - `name: "read"`

4075

4076 - `"read"`

4077

4078 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

4079

4080 Permission policy for tool execution.

4081

4082 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4083

4084 Tool calls are automatically approved without user confirmation.

4085

4086 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4087

4088 Tool calls require user confirmation before execution.

4089

4090 - `type: "read"`

4091

4092 - `"read"`

4093

4094 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

4095

4096 Configuration for the write tool.

4097

4098 - `enabled: boolean`

4099

4100 - `name: "write"`

4101

4102 - `"write"`

4103

4104 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

4105

4106 Permission policy for tool execution.

4107

4108 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4109

4110 Tool calls are automatically approved without user confirmation.

4111

4112 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4113

4114 Tool calls require user confirmation before execution.

4115

4116 - `type: "write"`

4117

4118 - `"write"`

4119

4120 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

4121

4122 Configuration for the glob tool.

4123

4124 - `enabled: boolean`

4125

4126 - `name: "glob"`

4127

4128 - `"glob"`

4129

4130 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

4131

4132 Permission policy for tool execution.

4133

4134 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4135

4136 Tool calls are automatically approved without user confirmation.

4137

4138 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4139

4140 Tool calls require user confirmation before execution.

4141

4142 - `type: "glob"`

4143

4144 - `"glob"`

4145

4146 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

4147

4148 Configuration for the grep tool.

4149

4150 - `enabled: boolean`

4151

4152 - `name: "grep"`

4153

4154 - `"grep"`

4155

4156 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

4157

4158 Permission policy for tool execution.

4159

4160 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4161

4162 Tool calls are automatically approved without user confirmation.

4163

4164 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4165

4166 Tool calls require user confirmation before execution.

4167

4168 - `type: "grep"`

4169

4170 - `"grep"`

4171

4172 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

4173

4174 Configuration for the web_fetch tool.

4175

4176 - `enabled: boolean`

4177

4178 - `name: "web_fetch"`

4179

4180 - `"web_fetch"`

4181

4182 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

4183

4184 Permission policy for tool execution.

4185

4186 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4187

4188 Tool calls are automatically approved without user confirmation.

4189

4190 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4191

4192 Tool calls require user confirmation before execution.

4193

4194 - `type: "web_fetch"`

4195

4196 - `"web_fetch"`

4197

4198 - `allowed_domains: optional array of string`

4199

4200 - `blocked_domains: optional array of string`

4201

4202 - `max_content_tokens: optional number or null`

4203

4204 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

4205

4206 Configuration for the web_search tool.

4207

4208 - `enabled: boolean`

4209

4210 - `name: "web_search"`

4211

4212 - `"web_search"`

4213

4214 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

4215

4216 Permission policy for tool execution.

4217

4218 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4219

4220 Tool calls are automatically approved without user confirmation.

4221

4222 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4223

4224 Tool calls require user confirmation before execution.

4225

4226 - `type: "web_search"`

4227

4228 - `"web_search"`

4229

4230 - `allowed_domains: optional array of string`

4231

4232 - `blocked_domains: optional array of string`

4233

4234 - `user_location: optional BetaManagedAgentsUserLocation or null`

4235

4236 Approximate user location for search result localization.

4237

4238 - `type: "approximate"`

4239

4240 Location precision. Only "approximate" is supported.

4241

4242 - `"approximate"`

4243

4244 - `city: optional string or null`

4245

4246 City name.

4247

4248 - `country: optional string or null`

4249

4250 Two-letter ISO 3166-1 country code, uppercase.

4251

4252 - `region: optional string or null`

4253

4254 Region or state name.

4255

4256 - `timezone: optional string or null`

4257

4258 IANA timezone identifier, e.g. "America/Los_Angeles".

29164259

29174260 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

29184261

from line 4432

30894432 "name": "bash",

30904433 "permission_policy": {

30914434 "type": "always_allow"

3092 }

4435 },

4436 "type": "bash"

30934437 }

30944438 ],

30954439 "default_config": {

from line 4767

34234767

34244768 - `configs: array of BetaManagedAgentsAgentToolConfig`

34254769

3426 - `enabled: boolean`

3427

3428 - `name: "bash" or "edit" or "read" or 5 more`

3429

3430 Built-in agent tool identifier.

3431

3432 - `"bash"`

3433

3434 - `"edit"`

3435

3436 - `"read"`

3437

3438 - `"write"`

3439

3440 - `"glob"`

3441

3442 - `"grep"`

3443

3444 - `"web_fetch"`

3445

3446 - `"web_search"`

3447

3448 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

3449

3450 Permission policy for tool execution.

3451

3452 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

3453

3454 Tool calls are automatically approved without user confirmation.

3455

3456 - `type: "always_allow"`

3457

3458 - `"always_allow"`

3459

3460 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

3461

3462 Tool calls require user confirmation before execution.

3463

3464 - `type: "always_ask"`

3465

3466 - `"always_ask"`

4770 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

4771

4772 Configuration for the bash tool.

4773

4774 - `enabled: boolean`

4775

4776 - `name: "bash"`

4777

4778 - `"bash"`

4779

4780 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

4781

4782 Permission policy for tool execution.

4783

4784 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4785

4786 Tool calls are automatically approved without user confirmation.

4787

4788 - `type: "always_allow"`

4789

4790 - `"always_allow"`

4791

4792 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4793

4794 Tool calls require user confirmation before execution.

4795

4796 - `type: "always_ask"`

4797

4798 - `"always_ask"`

4799

4800 - `type: "bash"`

4801

4802 - `"bash"`

4803

4804 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

4805

4806 Configuration for the edit tool.

4807

4808 - `enabled: boolean`

4809

4810 - `name: "edit"`

4811

4812 - `"edit"`

4813

4814 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

4815

4816 Permission policy for tool execution.

4817

4818 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4819

4820 Tool calls are automatically approved without user confirmation.

4821

4822 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4823

4824 Tool calls require user confirmation before execution.

4825

4826 - `type: "edit"`

4827

4828 - `"edit"`

4829

4830 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

4831

4832 Configuration for the read tool.

4833

4834 - `enabled: boolean`

4835

4836 - `name: "read"`

4837

4838 - `"read"`

4839

4840 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

4841

4842 Permission policy for tool execution.

4843

4844 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4845

4846 Tool calls are automatically approved without user confirmation.

4847

4848 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4849

4850 Tool calls require user confirmation before execution.

4851

4852 - `type: "read"`

4853

4854 - `"read"`

4855

4856 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

4857

4858 Configuration for the write tool.

4859

4860 - `enabled: boolean`

4861

4862 - `name: "write"`

4863

4864 - `"write"`

4865

4866 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

4867

4868 Permission policy for tool execution.

4869

4870 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4871

4872 Tool calls are automatically approved without user confirmation.

4873

4874 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4875

4876 Tool calls require user confirmation before execution.

4877

4878 - `type: "write"`

4879

4880 - `"write"`

4881

4882 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

4883

4884 Configuration for the glob tool.

4885

4886 - `enabled: boolean`

4887

4888 - `name: "glob"`

4889

4890 - `"glob"`

4891

4892 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

4893

4894 Permission policy for tool execution.

4895

4896 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4897

4898 Tool calls are automatically approved without user confirmation.

4899

4900 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4901

4902 Tool calls require user confirmation before execution.

4903

4904 - `type: "glob"`

4905

4906 - `"glob"`

4907

4908 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

4909

4910 Configuration for the grep tool.

4911

4912 - `enabled: boolean`

4913

4914 - `name: "grep"`

4915

4916 - `"grep"`

4917

4918 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

4919

4920 Permission policy for tool execution.

4921

4922 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4923

4924 Tool calls are automatically approved without user confirmation.

4925

4926 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4927

4928 Tool calls require user confirmation before execution.

4929

4930 - `type: "grep"`

4931

4932 - `"grep"`

4933

4934 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

4935

4936 Configuration for the web_fetch tool.

4937

4938 - `enabled: boolean`

4939

4940 - `name: "web_fetch"`

4941

4942 - `"web_fetch"`

4943

4944 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

4945

4946 Permission policy for tool execution.

4947

4948 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4949

4950 Tool calls are automatically approved without user confirmation.

4951

4952 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4953

4954 Tool calls require user confirmation before execution.

4955

4956 - `type: "web_fetch"`

4957

4958 - `"web_fetch"`

4959

4960 - `allowed_domains: optional array of string`

4961

4962 - `blocked_domains: optional array of string`

4963

4964 - `max_content_tokens: optional number or null`

4965

4966 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

4967

4968 Configuration for the web_search tool.

4969

4970 - `enabled: boolean`

4971

4972 - `name: "web_search"`

4973

4974 - `"web_search"`

4975

4976 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

4977

4978 Permission policy for tool execution.

4979

4980 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4981

4982 Tool calls are automatically approved without user confirmation.

4983

4984 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4985

4986 Tool calls require user confirmation before execution.

4987

4988 - `type: "web_search"`

4989

4990 - `"web_search"`

4991

4992 - `allowed_domains: optional array of string`

4993

4994 - `blocked_domains: optional array of string`

4995

4996 - `user_location: optional BetaManagedAgentsUserLocation or null`

4997

4998 Approximate user location for search result localization.

4999

5000 - `type: "approximate"`

5001

5002 Location precision. Only "approximate" is supported.

5003

5004 - `"approximate"`

5005

5006 - `city: optional string or null`

5007

5008 City name.

5009

5010 - `country: optional string or null`

5011

5012 Two-letter ISO 3166-1 country code, uppercase.

5013

5014 - `region: optional string or null`

5015

5016 Region or state name.

5017

5018 - `timezone: optional string or null`

5019

5020 IANA timezone identifier, e.g. "America/Los_Angeles".

34675021

34685022 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

34695023

from line 5189

36355189 "name": "bash",

36365190 "permission_policy": {

36375191 "type": "always_allow"

3638 }

5192 },

5193 "type": "bash"

36395194 }

36405195 ],

36415196 "default_config": {

from line 5452

38975452

38985453 - `configs: array of BetaManagedAgentsAgentToolConfig`

38995454

5455 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

5456

5457 Configuration for the bash tool.

5458

5459 - `enabled: boolean`

5460

5461 - `name: "bash"`

5462

5463 - `"bash"`

5464

5465 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

5466

5467 Permission policy for tool execution.

5468

5469 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

5470

5471 Tool calls are automatically approved without user confirmation.

5472

5473 - `type: "always_allow"`

5474

5475 - `"always_allow"`

5476

5477 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

5478

5479 Tool calls require user confirmation before execution.

5480

5481 - `type: "always_ask"`

5482

5483 - `"always_ask"`

5484

5485 - `type: "bash"`

5486

5487 - `"bash"`

5488

5489 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

5490

5491 Configuration for the edit tool.

5492

5493 - `enabled: boolean`

5494

5495 - `name: "edit"`

5496

5497 - `"edit"`

5498

5499 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

5500

5501 Permission policy for tool execution.

5502

5503 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

5504

5505 Tool calls are automatically approved without user confirmation.

5506

5507 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

5508

5509 Tool calls require user confirmation before execution.

5510

5511 - `type: "edit"`

5512

5513 - `"edit"`

5514

5515 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

5516

5517 Configuration for the read tool.

5518

5519 - `enabled: boolean`

5520

5521 - `name: "read"`

5522

5523 - `"read"`

5524

5525 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

5526

5527 Permission policy for tool execution.

5528

5529 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

5530

5531 Tool calls are automatically approved without user confirmation.

5532

5533 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

5534

5535 Tool calls require user confirmation before execution.

5536

5537 - `type: "read"`

5538

5539 - `"read"`

5540

5541 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

5542

5543 Configuration for the write tool.

5544

5545 - `enabled: boolean`

5546

5547 - `name: "write"`

5548

5549 - `"write"`

5550

5551 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

5552

5553 Permission policy for tool execution.

5554

5555 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

5556

5557 Tool calls are automatically approved without user confirmation.

5558

5559 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

5560

5561 Tool calls require user confirmation before execution.

5562

5563 - `type: "write"`

5564

5565 - `"write"`

5566

5567 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

5568

5569 Configuration for the glob tool.

5570

5571 - `enabled: boolean`

5572

5573 - `name: "glob"`

5574

5575 - `"glob"`

5576

5577 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

5578

5579 Permission policy for tool execution.

5580

5581 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

5582

5583 Tool calls are automatically approved without user confirmation.

5584

5585 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

5586

5587 Tool calls require user confirmation before execution.

5588

5589 - `type: "glob"`

5590

5591 - `"glob"`

5592

5593 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

5594

5595 Configuration for the grep tool.

5596

5597 - `enabled: boolean`

5598

5599 - `name: "grep"`

5600

5601 - `"grep"`

5602

5603 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

5604

5605 Permission policy for tool execution.

5606

5607 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

5608

5609 Tool calls are automatically approved without user confirmation.

5610

5611 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

5612

5613 Tool calls require user confirmation before execution.

5614

5615 - `type: "grep"`

5616

5617 - `"grep"`

5618

5619 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

5620

5621 Configuration for the web_fetch tool.

5622

5623 - `enabled: boolean`

5624

5625 - `name: "web_fetch"`

5626

5627 - `"web_fetch"`

5628

5629 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

5630

5631 Permission policy for tool execution.

5632

5633 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

5634

5635 Tool calls are automatically approved without user confirmation.

5636

5637 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

5638

5639 Tool calls require user confirmation before execution.

5640

5641 - `type: "web_fetch"`

5642

5643 - `"web_fetch"`

5644

5645 - `allowed_domains: optional array of string`

5646

5647 - `blocked_domains: optional array of string`

5648

5649 - `max_content_tokens: optional number or null`

5650

5651 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

5652

5653 Configuration for the web_search tool.

5654

5655 - `enabled: boolean`

5656

5657 - `name: "web_search"`

5658

5659 - `"web_search"`

5660

5661 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

5662

5663 Permission policy for tool execution.

5664

5665 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

5666

5667 Tool calls are automatically approved without user confirmation.

5668

5669 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

5670

5671 Tool calls require user confirmation before execution.

5672

5673 - `type: "web_search"`

5674

5675 - `"web_search"`

5676

5677 - `allowed_domains: optional array of string`

5678

5679 - `blocked_domains: optional array of string`

5680

5681 - `user_location: optional BetaManagedAgentsUserLocation or null`

5682

5683 Approximate user location for search result localization.

5684

5685 - `type: "approximate"`

5686

5687 Location precision. Only "approximate" is supported.

5688

5689 - `"approximate"`

5690

5691 - `city: optional string or null`

5692

5693 City name.

5694

5695 - `country: optional string or null`

5696

5697 Two-letter ISO 3166-1 country code, uppercase.

5698

5699 - `region: optional string or null`

5700

5701 Region or state name.

5702

5703 - `timezone: optional string or null`

5704

5705 IANA timezone identifier, e.g. "America/Los_Angeles".

5706

5707 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

5708

5709 Resolved default configuration for agent tools.

5710

39005711 - `enabled: boolean`

39015712

3902 - `name: "bash" or "edit" or "read" or 5 more`

3903

3904 Built-in agent tool identifier.

3905

3906 - `"bash"`

3907

3908 - `"edit"`

3909

3910 - `"read"`

3911

3912 - `"write"`

3913

3914 - `"glob"`

3915

3916 - `"grep"`

3917

3918 - `"web_fetch"`

3919

3920 - `"web_search"`

3921

39225713 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

39235714

39245715 Permission policy for tool execution.

from line 5718

39275718

39285719 Tool calls are automatically approved without user confirmation.

39295720

3930 - `type: "always_allow"`

3931

3932 - `"always_allow"`

3933

39345721 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

39355722

39365723 Tool calls require user confirmation before execution.

39375724

3938 - `type: "always_ask"`

3939

3940 - `"always_ask"`

3941

3942 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

3943

3944 Resolved default configuration for agent tools.

5725 - `type: "agent_toolset_20260401"`

5726

5727 - `"agent_toolset_20260401"`

5728

5729 - `BetaManagedAgentsMCPToolset object { configs, default_config, mcp_server_name, type }`

5730

5731 - `configs: array of BetaManagedAgentsMCPToolConfig`

39455732

39465733 - `enabled: boolean`

39475734

5735 - `name: string`

5736

39485737 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

39495738

39505739 Permission policy for tool execution.

from line 5746

39575746

39585747 Tool calls require user confirmation before execution.

39595748

3960 - `type: "agent_toolset_20260401"`

3961

3962 - `"agent_toolset_20260401"`

3963

3964 - `BetaManagedAgentsMCPToolset object { configs, default_config, mcp_server_name, type }`

3965

3966 - `configs: array of BetaManagedAgentsMCPToolConfig`

5749 - `default_config: BetaManagedAgentsMCPToolsetDefaultConfig`

5750

5751 Resolved default configuration for all tools from an MCP server.

39675752

39685753 - `enabled: boolean`

39695754

3970 - `name: string`

3971

39725755 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

39735756

39745757 Permission policy for tool execution.

from line 5764

39815764

39825765 Tool calls require user confirmation before execution.

39835766

3984 - `default_config: BetaManagedAgentsMCPToolsetDefaultConfig`

3985

3986 Resolved default configuration for all tools from an MCP server.

3987

3988 - `enabled: boolean`

3989

3990 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

3991

3992 Permission policy for tool execution.

3993

3994 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

3995

3996 Tool calls are automatically approved without user confirmation.

3997

3998 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

3999

4000 Tool calls require user confirmation before execution.

4001

40025767 - `mcp_server_name: string`

40035768

40045769 - `type: "mcp_toolset"`

from line 5822

40575822

40585823### Beta Managed Agents Agent Tool Config

40595824

4060- `BetaManagedAgentsAgentToolConfig object { enabled, name, permission_policy }`

5825- `BetaManagedAgentsAgentToolConfig = BetaManagedAgentsBashToolConfig or BetaManagedAgentsEditToolConfig or BetaManagedAgentsReadToolConfig or 5 more`

40615826

40625827 Configuration for a specific agent tool.

40635828

5829 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

5830

5831 Configuration for the bash tool.

5832

5833 - `enabled: boolean`

5834

5835 - `name: "bash"`

5836

5837 - `"bash"`

5838

5839 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

5840

5841 Permission policy for tool execution.

5842

5843 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

5844

5845 Tool calls are automatically approved without user confirmation.

5846

5847 - `type: "always_allow"`

5848

5849 - `"always_allow"`

5850

5851 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

5852

5853 Tool calls require user confirmation before execution.

5854

5855 - `type: "always_ask"`

5856

5857 - `"always_ask"`

5858

5859 - `type: "bash"`

5860

5861 - `"bash"`

5862

5863 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

5864

5865 Configuration for the edit tool.

5866

5867 - `enabled: boolean`

5868

5869 - `name: "edit"`

5870

5871 - `"edit"`

5872

5873 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

5874

5875 Permission policy for tool execution.

5876

5877 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

5878

5879 Tool calls are automatically approved without user confirmation.

5880

5881 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

5882

5883 Tool calls require user confirmation before execution.

5884

5885 - `type: "edit"`

5886

5887 - `"edit"`

5888

5889 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

5890

5891 Configuration for the read tool.

5892

5893 - `enabled: boolean`

5894

5895 - `name: "read"`

5896

5897 - `"read"`

5898

5899 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

5900

5901 Permission policy for tool execution.

5902

5903 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

5904

5905 Tool calls are automatically approved without user confirmation.

5906

5907 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

5908

5909 Tool calls require user confirmation before execution.

5910

5911 - `type: "read"`

5912

5913 - `"read"`

5914

5915 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

5916

5917 Configuration for the write tool.

5918

5919 - `enabled: boolean`

5920

5921 - `name: "write"`

5922

5923 - `"write"`

5924

5925 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

5926

5927 Permission policy for tool execution.

5928

5929 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

5930

5931 Tool calls are automatically approved without user confirmation.

5932

5933 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

5934

5935 Tool calls require user confirmation before execution.

5936

5937 - `type: "write"`

5938

5939 - `"write"`

5940

5941 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

5942

5943 Configuration for the glob tool.

5944

5945 - `enabled: boolean`

5946

5947 - `name: "glob"`

5948

5949 - `"glob"`

5950

5951 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

5952

5953 Permission policy for tool execution.

5954

5955 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

5956

5957 Tool calls are automatically approved without user confirmation.

5958

5959 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

5960

5961 Tool calls require user confirmation before execution.

5962

5963 - `type: "glob"`

5964

5965 - `"glob"`

5966

5967 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

5968

5969 Configuration for the grep tool.

5970

5971 - `enabled: boolean`

5972

5973 - `name: "grep"`

5974

5975 - `"grep"`

5976

5977 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

5978

5979 Permission policy for tool execution.

5980

5981 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

5982

5983 Tool calls are automatically approved without user confirmation.

5984

5985 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

5986

5987 Tool calls require user confirmation before execution.

5988

5989 - `type: "grep"`

5990

5991 - `"grep"`

5992

5993 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

5994

5995 Configuration for the web_fetch tool.

5996

5997 - `enabled: boolean`

5998

5999 - `name: "web_fetch"`

6000

6001 - `"web_fetch"`

6002

6003 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

6004

6005 Permission policy for tool execution.

6006

6007 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6008

6009 Tool calls are automatically approved without user confirmation.

6010

6011 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6012

6013 Tool calls require user confirmation before execution.

6014

6015 - `type: "web_fetch"`

6016

6017 - `"web_fetch"`

6018

6019 - `allowed_domains: optional array of string`

6020

6021 - `blocked_domains: optional array of string`

6022

6023 - `max_content_tokens: optional number or null`

6024

6025 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

6026

6027 Configuration for the web_search tool.

6028

6029 - `enabled: boolean`

6030

6031 - `name: "web_search"`

6032

6033 - `"web_search"`

6034

6035 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

6036

6037 Permission policy for tool execution.

6038

6039 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6040

6041 Tool calls are automatically approved without user confirmation.

6042

6043 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6044

6045 Tool calls require user confirmation before execution.

6046

6047 - `type: "web_search"`

6048

6049 - `"web_search"`

6050

6051 - `allowed_domains: optional array of string`

6052

6053 - `blocked_domains: optional array of string`

6054

6055 - `user_location: optional BetaManagedAgentsUserLocation or null`

6056

6057 Approximate user location for search result localization.

6058

6059 - `type: "approximate"`

6060

6061 Location precision. Only "approximate" is supported.

6062

6063 - `"approximate"`

6064

6065 - `city: optional string or null`

6066

6067 City name.

6068

6069 - `country: optional string or null`

6070

6071 Two-letter ISO 3166-1 country code, uppercase.

6072

6073 - `region: optional string or null`

6074

6075 Region or state name.

6076

6077 - `timezone: optional string or null`

6078

6079 IANA timezone identifier, e.g. "America/Los_Angeles".

6080

6081### Beta Managed Agents Agent Tool Config Params

6082

6083- `BetaManagedAgentsAgentToolConfigParams = BetaManagedAgentsBashToolConfigParams or BetaManagedAgentsEditToolConfigParams or BetaManagedAgentsReadToolConfigParams or 5 more`

6084

6085 Configuration override for a specific tool within a toolset.

6086

6087 - `BetaManagedAgentsBashToolConfigParams object { name, enabled, permission_policy, type }`

6088

6089 Configuration override for the bash tool.

6090

6091 - `name: "bash"`

6092

6093 Must be "bash".

6094

6095 - `"bash"`

6096

6097 - `enabled: optional boolean or null`

6098

6099 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

6100

6101 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

6102

6103 Permission policy for tool execution.

6104

6105 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6106

6107 Tool calls are automatically approved without user confirmation.

6108

6109 - `type: "always_allow"`

6110

6111 - `"always_allow"`

6112

6113 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6114

6115 Tool calls require user confirmation before execution.

6116

6117 - `type: "always_ask"`

6118

6119 - `"always_ask"`

6120

6121 - `type: optional "bash"`

6122

6123 - `"bash"`

6124

6125 - `BetaManagedAgentsEditToolConfigParams object { name, enabled, permission_policy, type }`

6126

6127 Configuration override for the edit tool.

6128

6129 - `name: "edit"`

6130

6131 Must be "edit".

6132

6133 - `"edit"`

6134

6135 - `enabled: optional boolean or null`

6136

6137 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

6138

6139 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

6140

6141 Permission policy for tool execution.

6142

6143 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6144

6145 Tool calls are automatically approved without user confirmation.

6146

6147 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6148

6149 Tool calls require user confirmation before execution.

6150

6151 - `type: optional "edit"`

6152

6153 - `"edit"`

6154

6155 - `BetaManagedAgentsReadToolConfigParams object { name, enabled, permission_policy, type }`

6156

6157 Configuration override for the read tool.

6158

6159 - `name: "read"`

6160

6161 Must be "read".

6162

6163 - `"read"`

6164

6165 - `enabled: optional boolean or null`

6166

6167 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

6168

6169 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

6170

6171 Permission policy for tool execution.

6172

6173 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6174

6175 Tool calls are automatically approved without user confirmation.

6176

6177 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6178

6179 Tool calls require user confirmation before execution.

6180

6181 - `type: optional "read"`

6182

6183 - `"read"`

6184

6185 - `BetaManagedAgentsWriteToolConfigParams object { name, enabled, permission_policy, type }`

6186

6187 Configuration override for the write tool.

6188

6189 - `name: "write"`

6190

6191 Must be "write".

6192

6193 - `"write"`

6194

6195 - `enabled: optional boolean or null`

6196

6197 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

6198

6199 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

6200

6201 Permission policy for tool execution.

6202

6203 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6204

6205 Tool calls are automatically approved without user confirmation.

6206

6207 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6208

6209 Tool calls require user confirmation before execution.

6210

6211 - `type: optional "write"`

6212

6213 - `"write"`

6214

6215 - `BetaManagedAgentsGlobToolConfigParams object { name, enabled, permission_policy, type }`

6216

6217 Configuration override for the glob tool.

6218

6219 - `name: "glob"`

6220

6221 Must be "glob".

6222

6223 - `"glob"`

6224

6225 - `enabled: optional boolean or null`

6226

6227 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

6228

6229 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

6230

6231 Permission policy for tool execution.

6232

6233 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6234

6235 Tool calls are automatically approved without user confirmation.

6236

6237 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6238

6239 Tool calls require user confirmation before execution.

6240

6241 - `type: optional "glob"`

6242

6243 - `"glob"`

6244

6245 - `BetaManagedAgentsGrepToolConfigParams object { name, enabled, permission_policy, type }`

6246

6247 Configuration override for the grep tool.

6248

6249 - `name: "grep"`

6250

6251 Must be "grep".

6252

6253 - `"grep"`

6254

6255 - `enabled: optional boolean or null`

6256

6257 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

6258

6259 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

6260

6261 Permission policy for tool execution.

6262

6263 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6264

6265 Tool calls are automatically approved without user confirmation.

6266

6267 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6268

6269 Tool calls require user confirmation before execution.

6270

6271 - `type: optional "grep"`

6272

6273 - `"grep"`

6274

6275 - `BetaManagedAgentsWebFetchToolConfigParams object { name, allowed_domains, blocked_domains, 4 more }`

6276

6277 Configuration override for the web_fetch tool.

6278

6279 - `name: "web_fetch"`

6280

6281 Must be "web_fetch".

6282

6283 - `"web_fetch"`

6284

6285 - `allowed_domains: optional array of string`

6286

6287 Only fetch URLs whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "docs.example.com" (no scheme, port, or path). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with blocked_domains.

6288

6289 - `blocked_domains: optional array of string`

6290

6291 Never fetch URLs whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "ads.example.com" (no scheme, port, or path). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with allowed_domains.

6292

6293 - `enabled: optional boolean or null`

6294

6295 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

6296

6297 - `max_content_tokens: optional number or null`

6298

6299 Maximum number of tokens of fetched text content to include in context per call. Does not apply to binary content such as PDFs.

6300

6301 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

6302

6303 Permission policy for tool execution.

6304

6305 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6306

6307 Tool calls are automatically approved without user confirmation.

6308

6309 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6310

6311 Tool calls require user confirmation before execution.

6312

6313 - `type: optional "web_fetch"`

6314

6315 - `"web_fetch"`

6316

6317 - `BetaManagedAgentsWebSearchToolConfigParams object { name, allowed_domains, blocked_domains, 4 more }`

6318

6319 Configuration override for the web_search tool.

6320

6321 - `name: "web_search"`

6322

6323 Must be "web_search".

6324

6325 - `"web_search"`

6326

6327 - `allowed_domains: optional array of string`

6328

6329 Only return search results whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "docs.example.com" (no scheme or port; an optional path suffix is accepted). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with blocked_domains.

6330

6331 - `blocked_domains: optional array of string`

6332

6333 Never return search results whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "ads.example.com" (no scheme or port; an optional path suffix is accepted). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with allowed_domains.

6334

6335 - `enabled: optional boolean or null`

6336

6337 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

6338

6339 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

6340

6341 Permission policy for tool execution.

6342

6343 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6344

6345 Tool calls are automatically approved without user confirmation.

6346

6347 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6348

6349 Tool calls require user confirmation before execution.

6350

6351 - `type: optional "web_search"`

6352

6353 - `"web_search"`

6354

6355 - `user_location: optional BetaManagedAgentsUserLocation or null`

6356

6357 Approximate user location for search result localization.

6358

6359 - `type: "approximate"`

6360

6361 Location precision. Only "approximate" is supported.

6362

6363 - `"approximate"`

6364

6365 - `city: optional string or null`

6366

6367 City name.

6368

6369 - `country: optional string or null`

6370

6371 Two-letter ISO 3166-1 country code, uppercase.

6372

6373 - `region: optional string or null`

6374

6375 Region or state name.

6376

6377 - `timezone: optional string or null`

6378

6379 IANA timezone identifier, e.g. "America/Los_Angeles".

6380

6381### Beta Managed Agents Agent Toolset Default Config

6382

6383- `BetaManagedAgentsAgentToolsetDefaultConfig object { enabled, permission_policy }`

6384

6385 Resolved default configuration for agent tools.

6386

40646387 - `enabled: boolean`

40656388

4066 - `name: "bash" or "edit" or "read" or 5 more`

4067

4068 Built-in agent tool identifier.

4069

4070 - `"bash"`

4071

4072 - `"edit"`

4073

4074 - `"read"`

4075

4076 - `"write"`

4077

4078 - `"glob"`

4079

4080 - `"grep"`

4081

4082 - `"web_fetch"`

4083

4084 - `"web_search"`

4085

40866389 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

40876390

40886391 Permission policy for tool execution.

from line 6406

41036406

41046407 - `"always_ask"`

41056408

4106### Beta Managed Agents Agent Tool Config Params

4107

4108- `BetaManagedAgentsAgentToolConfigParams object { name, enabled, permission_policy }`

4109

4110 Configuration override for a specific tool within a toolset.

4111

4112 - `name: "bash" or "edit" or "read" or 5 more`

4113

4114 Built-in agent tool identifier.

4115

4116 - `"bash"`

4117

4118 - `"edit"`

4119

4120 - `"read"`

4121

4122 - `"write"`

4123

4124 - `"glob"`

4125

4126 - `"grep"`

4127

4128 - `"web_fetch"`

4129

4130 - `"web_search"`

6409### Beta Managed Agents Agent Toolset Default Config Params

6410

6411- `BetaManagedAgentsAgentToolsetDefaultConfigParams object { enabled, permission_policy }`

6412

6413 Default configuration for all tools in a toolset.

41316414

41326415 - `enabled: optional boolean or null`

41336416

4134 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

6417 Whether tools are enabled and available to Claude by default. Defaults to true if not specified.

41356418

41366419 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

41376420

from line 6436

41536436

41546437 - `"always_ask"`

41556438

4156### Beta Managed Agents Agent Toolset Default Config

4157

4158- `BetaManagedAgentsAgentToolsetDefaultConfig object { enabled, permission_policy }`

4159

4160 Resolved default configuration for agent tools.

4161

4162 - `enabled: boolean`

4163

4164 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

4165

4166 Permission policy for tool execution.

4167

4168 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4169

4170 Tool calls are automatically approved without user confirmation.

4171

4172 - `type: "always_allow"`

4173

4174 - `"always_allow"`

4175

4176 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4177

4178 Tool calls require user confirmation before execution.

4179

4180 - `type: "always_ask"`

4181

4182 - `"always_ask"`

4183

4184### Beta Managed Agents Agent Toolset Default Config Params

4185

4186- `BetaManagedAgentsAgentToolsetDefaultConfigParams object { enabled, permission_policy }`

4187

4188 Default configuration for all tools in a toolset.

4189

4190 - `enabled: optional boolean or null`

4191

4192 Whether tools are enabled and available to Claude by default. Defaults to true if not specified.

4193

4194 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

4195

4196 Permission policy for tool execution.

4197

4198 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4199

4200 Tool calls are automatically approved without user confirmation.

4201

4202 - `type: "always_allow"`

4203

4204 - `"always_allow"`

4205

4206 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4207

4208 Tool calls require user confirmation before execution.

4209

4210 - `type: "always_ask"`

4211

4212 - `"always_ask"`

4213

42146439### Beta Managed Agents Agent Toolset20260401

42156440

42166441- `BetaManagedAgentsAgentToolset20260401 object { configs, default_config, type }`

42176442

42186443 - `configs: array of BetaManagedAgentsAgentToolConfig`

42196444

4220 - `enabled: boolean`

4221

4222 - `name: "bash" or "edit" or "read" or 5 more`

4223

4224 Built-in agent tool identifier.

4225

4226 - `"bash"`

4227

4228 - `"edit"`

4229

4230 - `"read"`

4231

4232 - `"write"`

4233

4234 - `"glob"`

4235

4236 - `"grep"`

4237

4238 - `"web_fetch"`

4239

4240 - `"web_search"`

4241

4242 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

4243

4244 Permission policy for tool execution.

4245

4246 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4247

4248 Tool calls are automatically approved without user confirmation.

4249

4250 - `type: "always_allow"`

4251

4252 - `"always_allow"`

4253

4254 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4255

4256 Tool calls require user confirmation before execution.

4257

4258 - `type: "always_ask"`

4259

4260 - `"always_ask"`

6445 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

6446

6447 Configuration for the bash tool.

6448

6449 - `enabled: boolean`

6450

6451 - `name: "bash"`

6452

6453 - `"bash"`

6454

6455 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

6456

6457 Permission policy for tool execution.

6458

6459 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6460

6461 Tool calls are automatically approved without user confirmation.

6462

6463 - `type: "always_allow"`

6464

6465 - `"always_allow"`

6466

6467 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6468

6469 Tool calls require user confirmation before execution.

6470

6471 - `type: "always_ask"`

6472

6473 - `"always_ask"`

6474

6475 - `type: "bash"`

6476

6477 - `"bash"`

6478

6479 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

6480

6481 Configuration for the edit tool.

6482

6483 - `enabled: boolean`

6484

6485 - `name: "edit"`

6486

6487 - `"edit"`

6488

6489 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

6490

6491 Permission policy for tool execution.

6492

6493 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6494

6495 Tool calls are automatically approved without user confirmation.

6496

6497 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6498

6499 Tool calls require user confirmation before execution.

6500

6501 - `type: "edit"`

6502

6503 - `"edit"`

6504

6505 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

6506

6507 Configuration for the read tool.

6508

6509 - `enabled: boolean`

6510

6511 - `name: "read"`

6512

6513 - `"read"`

6514

6515 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

6516

6517 Permission policy for tool execution.

6518

6519 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6520

6521 Tool calls are automatically approved without user confirmation.

6522

6523 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6524

6525 Tool calls require user confirmation before execution.

6526

6527 - `type: "read"`

6528

6529 - `"read"`

6530

6531 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

6532

6533 Configuration for the write tool.

6534

6535 - `enabled: boolean`

6536

6537 - `name: "write"`

6538

6539 - `"write"`

6540

6541 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

6542

6543 Permission policy for tool execution.

6544

6545 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6546

6547 Tool calls are automatically approved without user confirmation.

6548

6549 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6550

6551 Tool calls require user confirmation before execution.

6552

6553 - `type: "write"`

6554

6555 - `"write"`

6556

6557 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

6558

6559 Configuration for the glob tool.

6560

6561 - `enabled: boolean`

6562

6563 - `name: "glob"`

6564

6565 - `"glob"`

6566

6567 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

6568

6569 Permission policy for tool execution.

6570

6571 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6572

6573 Tool calls are automatically approved without user confirmation.

6574

6575 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6576

6577 Tool calls require user confirmation before execution.

6578

6579 - `type: "glob"`

6580

6581 - `"glob"`

6582

6583 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

6584

6585 Configuration for the grep tool.

6586

6587 - `enabled: boolean`

6588

6589 - `name: "grep"`

6590

6591 - `"grep"`

6592

6593 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

6594

6595 Permission policy for tool execution.

6596

6597 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6598

6599 Tool calls are automatically approved without user confirmation.

6600

6601 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6602

6603 Tool calls require user confirmation before execution.

6604

6605 - `type: "grep"`

6606

6607 - `"grep"`

6608

6609 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

6610

6611 Configuration for the web_fetch tool.

6612

6613 - `enabled: boolean`

6614

6615 - `name: "web_fetch"`

6616

6617 - `"web_fetch"`

6618

6619 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

6620

6621 Permission policy for tool execution.

6622

6623 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6624

6625 Tool calls are automatically approved without user confirmation.

6626

6627 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6628

6629 Tool calls require user confirmation before execution.

6630

6631 - `type: "web_fetch"`

6632

6633 - `"web_fetch"`

6634

6635 - `allowed_domains: optional array of string`

6636

6637 - `blocked_domains: optional array of string`

6638

6639 - `max_content_tokens: optional number or null`

6640

6641 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

6642

6643 Configuration for the web_search tool.

6644

6645 - `enabled: boolean`

6646

6647 - `name: "web_search"`

6648

6649 - `"web_search"`

6650

6651 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

6652

6653 Permission policy for tool execution.

6654

6655 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6656

6657 Tool calls are automatically approved without user confirmation.

6658

6659 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6660

6661 Tool calls require user confirmation before execution.

6662

6663 - `type: "web_search"`

6664

6665 - `"web_search"`

6666

6667 - `allowed_domains: optional array of string`

6668

6669 - `blocked_domains: optional array of string`

6670

6671 - `user_location: optional BetaManagedAgentsUserLocation or null`

6672

6673 Approximate user location for search result localization.

6674

6675 - `type: "approximate"`

6676

6677 Location precision. Only "approximate" is supported.

6678

6679 - `"approximate"`

6680

6681 - `city: optional string or null`

6682

6683 City name.

6684

6685 - `country: optional string or null`

6686

6687 Two-letter ISO 3166-1 country code, uppercase.

6688

6689 - `region: optional string or null`

6690

6691 Region or state name.

6692

6693 - `timezone: optional string or null`

6694

6695 IANA timezone identifier, e.g. "America/Los_Angeles".

42616696

42626697 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

42636698

from line 6813

43786813

43796814 Per-tool configuration overrides.

43806815

4381 - `name: "bash" or "edit" or "read" or 5 more`

4382

4383 Built-in agent tool identifier.

4384

4385 - `"bash"`

4386

4387 - `"edit"`

4388

4389 - `"read"`

4390

4391 - `"write"`

4392

4393 - `"glob"`

4394

4395 - `"grep"`

4396

4397 - `"web_fetch"`

4398

4399 - `"web_search"`

4400

4401 - `enabled: optional boolean or null`

4402

4403 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

4404

4405 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

4406

4407 Permission policy for tool execution.

4408

4409 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

4410

4411 Tool calls are automatically approved without user confirmation.

4412

4413 - `type: "always_allow"`

4414

4415 - `"always_allow"`

4416

4417 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

4418

4419 Tool calls require user confirmation before execution.

4420

4421 - `type: "always_ask"`

4422

4423 - `"always_ask"`

6816 - `BetaManagedAgentsBashToolConfigParams object { name, enabled, permission_policy, type }`

6817

6818 Configuration override for the bash tool.

6819

6820 - `name: "bash"`

6821

6822 Must be "bash".

6823

6824 - `"bash"`

6825

6826 - `enabled: optional boolean or null`

6827

6828 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

6829

6830 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

6831

6832 Permission policy for tool execution.

6833

6834 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6835

6836 Tool calls are automatically approved without user confirmation.

6837

6838 - `type: "always_allow"`

6839

6840 - `"always_allow"`

6841

6842 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6843

6844 Tool calls require user confirmation before execution.

6845

6846 - `type: "always_ask"`

6847

6848 - `"always_ask"`

6849

6850 - `type: optional "bash"`

6851

6852 - `"bash"`

6853

6854 - `BetaManagedAgentsEditToolConfigParams object { name, enabled, permission_policy, type }`

6855

6856 Configuration override for the edit tool.

6857

6858 - `name: "edit"`

6859

6860 Must be "edit".

6861

6862 - `"edit"`

6863

6864 - `enabled: optional boolean or null`

6865

6866 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

6867

6868 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

6869

6870 Permission policy for tool execution.

6871

6872 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6873

6874 Tool calls are automatically approved without user confirmation.

6875

6876 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6877

6878 Tool calls require user confirmation before execution.

6879

6880 - `type: optional "edit"`

6881

6882 - `"edit"`

6883

6884 - `BetaManagedAgentsReadToolConfigParams object { name, enabled, permission_policy, type }`

6885

6886 Configuration override for the read tool.

6887

6888 - `name: "read"`

6889

6890 Must be "read".

6891

6892 - `"read"`

6893

6894 - `enabled: optional boolean or null`

6895

6896 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

6897

6898 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

6899

6900 Permission policy for tool execution.

6901

6902 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6903

6904 Tool calls are automatically approved without user confirmation.

6905

6906 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6907

6908 Tool calls require user confirmation before execution.

6909

6910 - `type: optional "read"`

6911

6912 - `"read"`

6913

6914 - `BetaManagedAgentsWriteToolConfigParams object { name, enabled, permission_policy, type }`

6915

6916 Configuration override for the write tool.

6917

6918 - `name: "write"`

6919

6920 Must be "write".

6921

6922 - `"write"`

6923

6924 - `enabled: optional boolean or null`

6925

6926 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

6927

6928 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

6929

6930 Permission policy for tool execution.

6931

6932 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6933

6934 Tool calls are automatically approved without user confirmation.

6935

6936 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6937

6938 Tool calls require user confirmation before execution.

6939

6940 - `type: optional "write"`

6941

6942 - `"write"`

6943

6944 - `BetaManagedAgentsGlobToolConfigParams object { name, enabled, permission_policy, type }`

6945

6946 Configuration override for the glob tool.

6947

6948 - `name: "glob"`

6949

6950 Must be "glob".

6951

6952 - `"glob"`

6953

6954 - `enabled: optional boolean or null`

6955

6956 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

6957

6958 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

6959

6960 Permission policy for tool execution.

6961

6962 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6963

6964 Tool calls are automatically approved without user confirmation.

6965

6966 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6967

6968 Tool calls require user confirmation before execution.

6969

6970 - `type: optional "glob"`

6971

6972 - `"glob"`

6973

6974 - `BetaManagedAgentsGrepToolConfigParams object { name, enabled, permission_policy, type }`

6975

6976 Configuration override for the grep tool.

6977

6978 - `name: "grep"`

6979

6980 Must be "grep".

6981

6982 - `"grep"`

6983

6984 - `enabled: optional boolean or null`

6985

6986 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

6987

6988 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

6989

6990 Permission policy for tool execution.

6991

6992 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6993

6994 Tool calls are automatically approved without user confirmation.

6995

6996 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6997

6998 Tool calls require user confirmation before execution.

6999

7000 - `type: optional "grep"`

7001

7002 - `"grep"`

7003

7004 - `BetaManagedAgentsWebFetchToolConfigParams object { name, allowed_domains, blocked_domains, 4 more }`

7005

7006 Configuration override for the web_fetch tool.

7007

7008 - `name: "web_fetch"`

7009

7010 Must be "web_fetch".

7011

7012 - `"web_fetch"`

7013

7014 - `allowed_domains: optional array of string`

7015

7016 Only fetch URLs whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "docs.example.com" (no scheme, port, or path). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with blocked_domains.

7017

7018 - `blocked_domains: optional array of string`

7019

7020 Never fetch URLs whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "ads.example.com" (no scheme, port, or path). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with allowed_domains.

7021

7022 - `enabled: optional boolean or null`

7023

7024 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

7025

7026 - `max_content_tokens: optional number or null`

7027

7028 Maximum number of tokens of fetched text content to include in context per call. Does not apply to binary content such as PDFs.

7029

7030 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

7031

7032 Permission policy for tool execution.

7033

7034 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

7035

7036 Tool calls are automatically approved without user confirmation.

7037

7038 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

7039

7040 Tool calls require user confirmation before execution.

7041

7042 - `type: optional "web_fetch"`

7043

7044 - `"web_fetch"`

7045

7046 - `BetaManagedAgentsWebSearchToolConfigParams object { name, allowed_domains, blocked_domains, 4 more }`

7047

7048 Configuration override for the web_search tool.

7049

7050 - `name: "web_search"`

7051

7052 Must be "web_search".

7053

7054 - `"web_search"`

7055

7056 - `allowed_domains: optional array of string`

7057

7058 Only return search results whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "docs.example.com" (no scheme or port; an optional path suffix is accepted). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with blocked_domains.

7059

7060 - `blocked_domains: optional array of string`

7061

7062 Never return search results whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "ads.example.com" (no scheme or port; an optional path suffix is accepted). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with allowed_domains.

7063

7064 - `enabled: optional boolean or null`

7065

7066 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

7067

7068 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

7069

7070 Permission policy for tool execution.

7071

7072 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

7073

7074 Tool calls are automatically approved without user confirmation.

7075

7076 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

7077

7078 Tool calls require user confirmation before execution.

7079

7080 - `type: optional "web_search"`

7081

7082 - `"web_search"`

7083

7084 - `user_location: optional BetaManagedAgentsUserLocation or null`

7085

7086 Approximate user location for search result localization.

7087

7088 - `type: "approximate"`

7089

7090 Location precision. Only "approximate" is supported.

7091

7092 - `"approximate"`

7093

7094 - `city: optional string or null`

7095

7096 City name.

7097

7098 - `country: optional string or null`

7099

7100 Two-letter ISO 3166-1 country code, uppercase.

7101

7102 - `region: optional string or null`

7103

7104 Region or state name.

7105

7106 - `timezone: optional string or null`

7107

7108 IANA timezone identifier, e.g. "America/Los_Angeles".

44247109

44257110 - `default_config: optional BetaManagedAgentsAgentToolsetDefaultConfigParams or null`

44267111

from line 7212

45277212

45287213 Version to pin. Defaults to latest if omitted.

45297214

7215### Beta Managed Agents Bash Tool Config

7216

7217- `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

7218

7219 Configuration for the bash tool.

7220

7221 - `enabled: boolean`

7222

7223 - `name: "bash"`

7224

7225 - `"bash"`

7226

7227 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

7228

7229 Permission policy for tool execution.

7230

7231 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

7232

7233 Tool calls are automatically approved without user confirmation.

7234

7235 - `type: "always_allow"`

7236

7237 - `"always_allow"`

7238

7239 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

7240

7241 Tool calls require user confirmation before execution.

7242

7243 - `type: "always_ask"`

7244

7245 - `"always_ask"`

7246

7247 - `type: "bash"`

7248

7249 - `"bash"`

7250

7251### Beta Managed Agents Bash Tool Config Params

7252

7253- `BetaManagedAgentsBashToolConfigParams object { name, enabled, permission_policy, type }`

7254

7255 Configuration override for the bash tool.

7256

7257 - `name: "bash"`

7258

7259 Must be "bash".

7260

7261 - `"bash"`

7262

7263 - `enabled: optional boolean or null`

7264

7265 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

7266

7267 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

7268

7269 Permission policy for tool execution.

7270

7271 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

7272

7273 Tool calls are automatically approved without user confirmation.

7274

7275 - `type: "always_allow"`

7276

7277 - `"always_allow"`

7278

7279 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

7280

7281 Tool calls require user confirmation before execution.

7282

7283 - `type: "always_ask"`

7284

7285 - `"always_ask"`

7286

7287 - `type: optional "bash"`

7288

7289 - `"bash"`

7290

45307291### Beta Managed Agents Custom Skill

45317292

45327293- `BetaManagedAgentsCustomSkill object { skill_id, type, version }`

from line 7390

46297390

46307391 - `"custom"`

46317392

7393### Beta Managed Agents Edit Tool Config

7394

7395- `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

7396

7397 Configuration for the edit tool.

7398

7399 - `enabled: boolean`

7400

7401 - `name: "edit"`

7402

7403 - `"edit"`

7404

7405 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

7406

7407 Permission policy for tool execution.

7408

7409 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

7410

7411 Tool calls are automatically approved without user confirmation.

7412

7413 - `type: "always_allow"`

7414

7415 - `"always_allow"`

7416

7417 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

7418

7419 Tool calls require user confirmation before execution.

7420

7421 - `type: "always_ask"`

7422

7423 - `"always_ask"`

7424

7425 - `type: "edit"`

7426

7427 - `"edit"`

7428

7429### Beta Managed Agents Edit Tool Config Params

7430

7431- `BetaManagedAgentsEditToolConfigParams object { name, enabled, permission_policy, type }`

7432

7433 Configuration override for the edit tool.

7434

7435 - `name: "edit"`

7436

7437 Must be "edit".

7438

7439 - `"edit"`

7440

7441 - `enabled: optional boolean or null`

7442

7443 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

7444

7445 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

7446

7447 Permission policy for tool execution.

7448

7449 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

7450

7451 Tool calls are automatically approved without user confirmation.

7452

7453 - `type: "always_allow"`

7454

7455 - `"always_allow"`

7456

7457 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

7458

7459 Tool calls require user confirmation before execution.

7460

7461 - `type: "always_ask"`

7462

7463 - `"always_ask"`

7464

7465 - `type: optional "edit"`

7466

7467 - `"edit"`

7468

46327469### Beta Managed Agents Effort High

46337470

46347471- `BetaManagedAgentsEffortHigh object { type }`

from line 7516

46797516

46807517 - `"xhigh"`

46817518

7519### Beta Managed Agents Glob Tool Config

7520

7521- `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

7522

7523 Configuration for the glob tool.

7524

7525 - `enabled: boolean`

7526

7527 - `name: "glob"`

7528

7529 - `"glob"`

7530

7531 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

7532

7533 Permission policy for tool execution.

7534

7535 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

7536

7537 Tool calls are automatically approved without user confirmation.

7538

7539 - `type: "always_allow"`

7540

7541 - `"always_allow"`

7542

7543 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

7544

7545 Tool calls require user confirmation before execution.

7546

7547 - `type: "always_ask"`

7548

7549 - `"always_ask"`

7550

7551 - `type: "glob"`

7552

7553 - `"glob"`

7554

7555### Beta Managed Agents Glob Tool Config Params

7556

7557- `BetaManagedAgentsGlobToolConfigParams object { name, enabled, permission_policy, type }`

7558

7559 Configuration override for the glob tool.

7560

7561 - `name: "glob"`

7562

7563 Must be "glob".

7564

7565 - `"glob"`

7566

7567 - `enabled: optional boolean or null`

7568

7569 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

7570

7571 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

7572

7573 Permission policy for tool execution.

7574

7575 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

7576

7577 Tool calls are automatically approved without user confirmation.

7578

7579 - `type: "always_allow"`

7580

7581 - `"always_allow"`

7582

7583 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

7584

7585 Tool calls require user confirmation before execution.

7586

7587 - `type: "always_ask"`

7588

7589 - `"always_ask"`

7590

7591 - `type: optional "glob"`

7592

7593 - `"glob"`

7594

7595### Beta Managed Agents Grep Tool Config

7596

7597- `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

7598

7599 Configuration for the grep tool.

7600

7601 - `enabled: boolean`

7602

7603 - `name: "grep"`

7604

7605 - `"grep"`

7606

7607 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

7608

7609 Permission policy for tool execution.

7610

7611 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

7612

7613 Tool calls are automatically approved without user confirmation.

7614

7615 - `type: "always_allow"`

7616

7617 - `"always_allow"`

7618

7619 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

7620

7621 Tool calls require user confirmation before execution.

7622

7623 - `type: "always_ask"`

7624

7625 - `"always_ask"`

7626

7627 - `type: "grep"`

7628

7629 - `"grep"`

7630

7631### Beta Managed Agents Grep Tool Config Params

7632

7633- `BetaManagedAgentsGrepToolConfigParams object { name, enabled, permission_policy, type }`

7634

7635 Configuration override for the grep tool.

7636

7637 - `name: "grep"`

7638

7639 Must be "grep".

7640

7641 - `"grep"`

7642

7643 - `enabled: optional boolean or null`

7644

7645 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

7646

7647 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

7648

7649 Permission policy for tool execution.

7650

7651 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

7652

7653 Tool calls are automatically approved without user confirmation.

7654

7655 - `type: "always_allow"`

7656

7657 - `"always_allow"`

7658

7659 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

7660

7661 Tool calls require user confirmation before execution.

7662

7663 - `type: "always_ask"`

7664

7665 - `"always_ask"`

7666

7667 - `type: optional "grep"`

7668

7669 - `"grep"`

7670

46827671### Beta Managed Agents MCP Server URL Definition

46837672

46847673- `BetaManagedAgentsMCPServerURLDefinition object { name, type, url }`

from line 8362

53738362

53748363 - `"self"`

53758364

8365### Beta Managed Agents Read Tool Config

8366

8367- `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

8368

8369 Configuration for the read tool.

8370

8371 - `enabled: boolean`

8372

8373 - `name: "read"`

8374

8375 - `"read"`

8376

8377 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

8378

8379 Permission policy for tool execution.

8380

8381 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

8382

8383 Tool calls are automatically approved without user confirmation.

8384

8385 - `type: "always_allow"`

8386

8387 - `"always_allow"`

8388

8389 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

8390

8391 Tool calls require user confirmation before execution.

8392

8393 - `type: "always_ask"`

8394

8395 - `"always_ask"`

8396

8397 - `type: "read"`

8398

8399 - `"read"`

8400

8401### Beta Managed Agents Read Tool Config Params

8402

8403- `BetaManagedAgentsReadToolConfigParams object { name, enabled, permission_policy, type }`

8404

8405 Configuration override for the read tool.

8406

8407 - `name: "read"`

8408

8409 Must be "read".

8410

8411 - `"read"`

8412

8413 - `enabled: optional boolean or null`

8414

8415 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

8416

8417 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

8418

8419 Permission policy for tool execution.

8420

8421 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

8422

8423 Tool calls are automatically approved without user confirmation.

8424

8425 - `type: "always_allow"`

8426

8427 - `"always_allow"`

8428

8429 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

8430

8431 Tool calls require user confirmation before execution.

8432

8433 - `type: "always_ask"`

8434

8435 - `"always_ask"`

8436

8437 - `type: optional "read"`

8438

8439 - `"read"`

8440

53768441### Beta Managed Agents Session Thread Agent

53778442

53788443- `BetaManagedAgentsSessionThreadAgent object { id, description, mcp_servers, 7 more }`

from line 8620

55558620

55568621 - `configs: array of BetaManagedAgentsAgentToolConfig`

55578622

8623 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

8624

8625 Configuration for the bash tool.

8626

8627 - `enabled: boolean`

8628

8629 - `name: "bash"`

8630

8631 - `"bash"`

8632

8633 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

8634

8635 Permission policy for tool execution.

8636

8637 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

8638

8639 Tool calls are automatically approved without user confirmation.

8640

8641 - `type: "always_allow"`

8642

8643 - `"always_allow"`

8644

8645 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

8646

8647 Tool calls require user confirmation before execution.

8648

8649 - `type: "always_ask"`

8650

8651 - `"always_ask"`

8652

8653 - `type: "bash"`

8654

8655 - `"bash"`

8656

8657 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

8658

8659 Configuration for the edit tool.

8660

8661 - `enabled: boolean`

8662

8663 - `name: "edit"`

8664

8665 - `"edit"`

8666

8667 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

8668

8669 Permission policy for tool execution.

8670

8671 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

8672

8673 Tool calls are automatically approved without user confirmation.

8674

8675 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

8676

8677 Tool calls require user confirmation before execution.

8678

8679 - `type: "edit"`

8680

8681 - `"edit"`

8682

8683 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

8684

8685 Configuration for the read tool.

8686

8687 - `enabled: boolean`

8688

8689 - `name: "read"`

8690

8691 - `"read"`

8692

8693 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

8694

8695 Permission policy for tool execution.

8696

8697 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

8698

8699 Tool calls are automatically approved without user confirmation.

8700

8701 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

8702

8703 Tool calls require user confirmation before execution.

8704

8705 - `type: "read"`

8706

8707 - `"read"`

8708

8709 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

8710

8711 Configuration for the write tool.

8712

8713 - `enabled: boolean`

8714

8715 - `name: "write"`

8716

8717 - `"write"`

8718

8719 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

8720

8721 Permission policy for tool execution.

8722

8723 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

8724

8725 Tool calls are automatically approved without user confirmation.

8726

8727 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

8728

8729 Tool calls require user confirmation before execution.

8730

8731 - `type: "write"`

8732

8733 - `"write"`

8734

8735 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

8736

8737 Configuration for the glob tool.

8738

8739 - `enabled: boolean`

8740

8741 - `name: "glob"`

8742

8743 - `"glob"`

8744

8745 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

8746

8747 Permission policy for tool execution.

8748

8749 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

8750

8751 Tool calls are automatically approved without user confirmation.

8752

8753 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

8754

8755 Tool calls require user confirmation before execution.

8756

8757 - `type: "glob"`

8758

8759 - `"glob"`

8760

8761 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

8762

8763 Configuration for the grep tool.

8764

8765 - `enabled: boolean`

8766

8767 - `name: "grep"`

8768

8769 - `"grep"`

8770

8771 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

8772

8773 Permission policy for tool execution.

8774

8775 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

8776

8777 Tool calls are automatically approved without user confirmation.

8778

8779 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

8780

8781 Tool calls require user confirmation before execution.

8782

8783 - `type: "grep"`

8784

8785 - `"grep"`

8786

8787 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

8788

8789 Configuration for the web_fetch tool.

8790

8791 - `enabled: boolean`

8792

8793 - `name: "web_fetch"`

8794

8795 - `"web_fetch"`

8796

8797 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

8798

8799 Permission policy for tool execution.

8800

8801 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

8802

8803 Tool calls are automatically approved without user confirmation.

8804

8805 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

8806

8807 Tool calls require user confirmation before execution.

8808

8809 - `type: "web_fetch"`

8810

8811 - `"web_fetch"`

8812

8813 - `allowed_domains: optional array of string`

8814

8815 - `blocked_domains: optional array of string`

8816

8817 - `max_content_tokens: optional number or null`

8818

8819 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

8820

8821 Configuration for the web_search tool.

8822

8823 - `enabled: boolean`

8824

8825 - `name: "web_search"`

8826

8827 - `"web_search"`

8828

8829 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

8830

8831 Permission policy for tool execution.

8832

8833 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

8834

8835 Tool calls are automatically approved without user confirmation.

8836

8837 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

8838

8839 Tool calls require user confirmation before execution.

8840

8841 - `type: "web_search"`

8842

8843 - `"web_search"`

8844

8845 - `allowed_domains: optional array of string`

8846

8847 - `blocked_domains: optional array of string`

8848

8849 - `user_location: optional BetaManagedAgentsUserLocation or null`

8850

8851 Approximate user location for search result localization.

8852

8853 - `type: "approximate"`

8854

8855 Location precision. Only "approximate" is supported.

8856

8857 - `"approximate"`

8858

8859 - `city: optional string or null`

8860

8861 City name.

8862

8863 - `country: optional string or null`

8864

8865 Two-letter ISO 3166-1 country code, uppercase.

8866

8867 - `region: optional string or null`

8868

8869 Region or state name.

8870

8871 - `timezone: optional string or null`

8872

8873 IANA timezone identifier, e.g. "America/Los_Angeles".

8874

8875 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

8876

8877 Resolved default configuration for agent tools.

8878

55588879 - `enabled: boolean`

55598880

5560 - `name: "bash" or "edit" or "read" or 5 more`

5561

5562 Built-in agent tool identifier.

5563

5564 - `"bash"`

5565

5566 - `"edit"`

5567

5568 - `"read"`

5569

5570 - `"write"`

5571

5572 - `"glob"`

5573

5574 - `"grep"`

5575

5576 - `"web_fetch"`

5577

5578 - `"web_search"`

5579

55808881 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

55818882

55828883 Permission policy for tool execution.

from line 8886

55858886

55868887 Tool calls are automatically approved without user confirmation.

55878888

5588 - `type: "always_allow"`

5589

5590 - `"always_allow"`

5591

55928889 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

55938890

55948891 Tool calls require user confirmation before execution.

55958892

5596 - `type: "always_ask"`

5597

5598 - `"always_ask"`

5599

5600 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

5601

5602 Resolved default configuration for agent tools.

8893 - `type: "agent_toolset_20260401"`

8894

8895 - `"agent_toolset_20260401"`

8896

8897 - `BetaManagedAgentsMCPToolset object { configs, default_config, mcp_server_name, type }`

8898

8899 - `configs: array of BetaManagedAgentsMCPToolConfig`

56038900

56048901 - `enabled: boolean`

56058902

8903 - `name: string`

8904

56068905 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

56078906

56088907 Permission policy for tool execution.

from line 8914

56158914

56168915 Tool calls require user confirmation before execution.

56178916

5618 - `type: "agent_toolset_20260401"`

5619

5620 - `"agent_toolset_20260401"`

5621

5622 - `BetaManagedAgentsMCPToolset object { configs, default_config, mcp_server_name, type }`

5623

5624 - `configs: array of BetaManagedAgentsMCPToolConfig`

8917 - `default_config: BetaManagedAgentsMCPToolsetDefaultConfig`

8918

8919 Resolved default configuration for all tools from an MCP server.

56258920

56268921 - `enabled: boolean`

56278922

5628 - `name: string`

5629

56308923 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

56318924

56328925 Permission policy for tool execution.

from line 8932

56398932

56408933 Tool calls require user confirmation before execution.

56418934

5642 - `default_config: BetaManagedAgentsMCPToolsetDefaultConfig`

5643

5644 Resolved default configuration for all tools from an MCP server.

5645

5646 - `enabled: boolean`

5647

5648 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

5649

5650 Permission policy for tool execution.

5651

5652 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

5653

5654 Tool calls are automatically approved without user confirmation.

5655

5656 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

5657

5658 Tool calls require user confirmation before execution.

5659

56608935 - `mcp_server_name: string`

56618936

56628937 - `type: "mcp_toolset"`

from line 9024

57499024

57509025 Endpoint URL for the MCP server.

57519026

9027### Beta Managed Agents User Location

9028

9029- `BetaManagedAgentsUserLocation object { type, city, country, 2 more }`

9030

9031 Approximate user location for search result localization.

9032

9033 - `type: "approximate"`

9034

9035 Location precision. Only "approximate" is supported.

9036

9037 - `"approximate"`

9038

9039 - `city: optional string or null`

9040

9041 City name.

9042

9043 - `country: optional string or null`

9044

9045 Two-letter ISO 3166-1 country code, uppercase.

9046

9047 - `region: optional string or null`

9048

9049 Region or state name.

9050

9051 - `timezone: optional string or null`

9052

9053 IANA timezone identifier, e.g. "America/Los_Angeles".

9054

9055### Beta Managed Agents Web Fetch Tool Config

9056

9057- `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

9058

9059 Configuration for the web_fetch tool.

9060

9061 - `enabled: boolean`

9062

9063 - `name: "web_fetch"`

9064

9065 - `"web_fetch"`

9066

9067 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

9068

9069 Permission policy for tool execution.

9070

9071 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

9072

9073 Tool calls are automatically approved without user confirmation.

9074

9075 - `type: "always_allow"`

9076

9077 - `"always_allow"`

9078

9079 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

9080

9081 Tool calls require user confirmation before execution.

9082

9083 - `type: "always_ask"`

9084

9085 - `"always_ask"`

9086

9087 - `type: "web_fetch"`

9088

9089 - `"web_fetch"`

9090

9091 - `allowed_domains: optional array of string`

9092

9093 - `blocked_domains: optional array of string`

9094

9095 - `max_content_tokens: optional number or null`

9096

9097### Beta Managed Agents Web Fetch Tool Config Params

9098

9099- `BetaManagedAgentsWebFetchToolConfigParams object { name, allowed_domains, blocked_domains, 4 more }`

9100

9101 Configuration override for the web_fetch tool.

9102

9103 - `name: "web_fetch"`

9104

9105 Must be "web_fetch".

9106

9107 - `"web_fetch"`

9108

9109 - `allowed_domains: optional array of string`

9110

9111 Only fetch URLs whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "docs.example.com" (no scheme, port, or path). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with blocked_domains.

9112

9113 - `blocked_domains: optional array of string`

9114

9115 Never fetch URLs whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "ads.example.com" (no scheme, port, or path). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with allowed_domains.

9116

9117 - `enabled: optional boolean or null`

9118

9119 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

9120

9121 - `max_content_tokens: optional number or null`

9122

9123 Maximum number of tokens of fetched text content to include in context per call. Does not apply to binary content such as PDFs.

9124

9125 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

9126

9127 Permission policy for tool execution.

9128

9129 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

9130

9131 Tool calls are automatically approved without user confirmation.

9132

9133 - `type: "always_allow"`

9134

9135 - `"always_allow"`

9136

9137 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

9138

9139 Tool calls require user confirmation before execution.

9140

9141 - `type: "always_ask"`

9142

9143 - `"always_ask"`

9144

9145 - `type: optional "web_fetch"`

9146

9147 - `"web_fetch"`

9148

9149### Beta Managed Agents Web Search Tool Config

9150

9151- `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

9152

9153 Configuration for the web_search tool.

9154

9155 - `enabled: boolean`

9156

9157 - `name: "web_search"`

9158

9159 - `"web_search"`

9160

9161 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

9162

9163 Permission policy for tool execution.

9164

9165 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

9166

9167 Tool calls are automatically approved without user confirmation.

9168

9169 - `type: "always_allow"`

9170

9171 - `"always_allow"`

9172

9173 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

9174

9175 Tool calls require user confirmation before execution.

9176

9177 - `type: "always_ask"`

9178

9179 - `"always_ask"`

9180

9181 - `type: "web_search"`

9182

9183 - `"web_search"`

9184

9185 - `allowed_domains: optional array of string`

9186

9187 - `blocked_domains: optional array of string`

9188

9189 - `user_location: optional BetaManagedAgentsUserLocation or null`

9190

9191 Approximate user location for search result localization.

9192

9193 - `type: "approximate"`

9194

9195 Location precision. Only "approximate" is supported.

9196

9197 - `"approximate"`

9198

9199 - `city: optional string or null`

9200

9201 City name.

9202

9203 - `country: optional string or null`

9204

9205 Two-letter ISO 3166-1 country code, uppercase.

9206

9207 - `region: optional string or null`

9208

9209 Region or state name.

9210

9211 - `timezone: optional string or null`

9212

9213 IANA timezone identifier, e.g. "America/Los_Angeles".

9214

9215### Beta Managed Agents Web Search Tool Config Params

9216

9217- `BetaManagedAgentsWebSearchToolConfigParams object { name, allowed_domains, blocked_domains, 4 more }`

9218

9219 Configuration override for the web_search tool.

9220

9221 - `name: "web_search"`

9222

9223 Must be "web_search".

9224

9225 - `"web_search"`

9226

9227 - `allowed_domains: optional array of string`

9228

9229 Only return search results whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "docs.example.com" (no scheme or port; an optional path suffix is accepted). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with blocked_domains.

9230

9231 - `blocked_domains: optional array of string`

9232

9233 Never return search results whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "ads.example.com" (no scheme or port; an optional path suffix is accepted). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with allowed_domains.

9234

9235 - `enabled: optional boolean or null`

9236

9237 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

9238

9239 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

9240

9241 Permission policy for tool execution.

9242

9243 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

9244

9245 Tool calls are automatically approved without user confirmation.

9246

9247 - `type: "always_allow"`

9248

9249 - `"always_allow"`

9250

9251 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

9252

9253 Tool calls require user confirmation before execution.

9254

9255 - `type: "always_ask"`

9256

9257 - `"always_ask"`

9258

9259 - `type: optional "web_search"`

9260

9261 - `"web_search"`

9262

9263 - `user_location: optional BetaManagedAgentsUserLocation or null`

9264

9265 Approximate user location for search result localization.

9266

9267 - `type: "approximate"`

9268

9269 Location precision. Only "approximate" is supported.

9270

9271 - `"approximate"`

9272

9273 - `city: optional string or null`

9274

9275 City name.

9276

9277 - `country: optional string or null`

9278

9279 Two-letter ISO 3166-1 country code, uppercase.

9280

9281 - `region: optional string or null`

9282

9283 Region or state name.

9284

9285 - `timezone: optional string or null`

9286

9287 IANA timezone identifier, e.g. "America/Los_Angeles".

9288

9289### Beta Managed Agents Write Tool Config

9290

9291- `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

9292

9293 Configuration for the write tool.

9294

9295 - `enabled: boolean`

9296

9297 - `name: "write"`

9298

9299 - `"write"`

9300

9301 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

9302

9303 Permission policy for tool execution.

9304

9305 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

9306

9307 Tool calls are automatically approved without user confirmation.

9308

9309 - `type: "always_allow"`

9310

9311 - `"always_allow"`

9312

9313 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

9314

9315 Tool calls require user confirmation before execution.

9316

9317 - `type: "always_ask"`

9318

9319 - `"always_ask"`

9320

9321 - `type: "write"`

9322

9323 - `"write"`

9324

9325### Beta Managed Agents Write Tool Config Params

9326

9327- `BetaManagedAgentsWriteToolConfigParams object { name, enabled, permission_policy, type }`

9328

9329 Configuration override for the write tool.

9330

9331 - `name: "write"`

9332

9333 Must be "write".

9334

9335 - `"write"`

9336

9337 - `enabled: optional boolean or null`

9338

9339 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

9340

9341 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

9342

9343 Permission policy for tool execution.

9344

9345 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

9346

9347 Tool calls are automatically approved without user confirmation.

9348

9349 - `type: "always_allow"`

9350

9351 - `"always_allow"`

9352

9353 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

9354

9355 Tool calls require user confirmation before execution.

9356

9357 - `type: "always_ask"`

9358

9359 - `"always_ask"`

9360

9361 - `type: optional "write"`

9362

9363 - `"write"`

9364

57529365# Versions

57539366

57549367## List Agent Versions

from line 9690

60779690

60789691 - `configs: array of BetaManagedAgentsAgentToolConfig`

60799692

6080 - `enabled: boolean`

6081

6082 - `name: "bash" or "edit" or "read" or 5 more`

6083

6084 Built-in agent tool identifier.

6085

6086 - `"bash"`

6087

6088 - `"edit"`

6089

6090 - `"read"`

6091

6092 - `"write"`

6093

6094 - `"glob"`

6095

6096 - `"grep"`

6097

6098 - `"web_fetch"`

6099

6100 - `"web_search"`

6101

6102 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

6103

6104 Permission policy for tool execution.

6105

6106 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

6107

6108 Tool calls are automatically approved without user confirmation.

6109

6110 - `type: "always_allow"`

6111

6112 - `"always_allow"`

6113

6114 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

6115

6116 Tool calls require user confirmation before execution.

6117

6118 - `type: "always_ask"`

6119

6120 - `"always_ask"`

9693 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

9694

9695 Configuration for the bash tool.

9696

9697 - `enabled: boolean`

9698

9699 - `name: "bash"`

9700

9701 - `"bash"`

9702

9703 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

9704

9705 Permission policy for tool execution.

9706

9707 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

9708

9709 Tool calls are automatically approved without user confirmation.

9710

9711 - `type: "always_allow"`

9712

9713 - `"always_allow"`

9714

9715 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

9716

9717 Tool calls require user confirmation before execution.

9718

9719 - `type: "always_ask"`

9720

9721 - `"always_ask"`

9722

9723 - `type: "bash"`

9724

9725 - `"bash"`

9726

9727 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

9728

9729 Configuration for the edit tool.

9730

9731 - `enabled: boolean`

9732

9733 - `name: "edit"`

9734

9735 - `"edit"`

9736

9737 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

9738

9739 Permission policy for tool execution.

9740

9741 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

9742

9743 Tool calls are automatically approved without user confirmation.

9744

9745 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

9746

9747 Tool calls require user confirmation before execution.

9748

9749 - `type: "edit"`

9750

9751 - `"edit"`

9752

9753 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

9754

9755 Configuration for the read tool.

9756

9757 - `enabled: boolean`

9758

9759 - `name: "read"`

9760

9761 - `"read"`

9762

9763 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

9764

9765 Permission policy for tool execution.

9766

9767 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

9768

9769 Tool calls are automatically approved without user confirmation.

9770

9771 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

9772

9773 Tool calls require user confirmation before execution.

9774

9775 - `type: "read"`

9776

9777 - `"read"`

9778

9779 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

9780

9781 Configuration for the write tool.

9782

9783 - `enabled: boolean`

9784

9785 - `name: "write"`

9786

9787 - `"write"`

9788

9789 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

9790

9791 Permission policy for tool execution.

9792

9793 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

9794

9795 Tool calls are automatically approved without user confirmation.

9796

9797 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

9798

9799 Tool calls require user confirmation before execution.

9800

9801 - `type: "write"`

9802

9803 - `"write"`

9804

9805 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

9806

9807 Configuration for the glob tool.

9808

9809 - `enabled: boolean`

9810

9811 - `name: "glob"`

9812

9813 - `"glob"`

9814

9815 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

9816

9817 Permission policy for tool execution.

9818

9819 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

9820

9821 Tool calls are automatically approved without user confirmation.

9822

9823 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

9824

9825 Tool calls require user confirmation before execution.

9826

9827 - `type: "glob"`

9828

9829 - `"glob"`

9830

9831 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

9832

9833 Configuration for the grep tool.

9834

9835 - `enabled: boolean`

9836

9837 - `name: "grep"`

9838

9839 - `"grep"`

9840

9841 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

9842

9843 Permission policy for tool execution.

9844

9845 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

9846

9847 Tool calls are automatically approved without user confirmation.

9848

9849 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

9850

9851 Tool calls require user confirmation before execution.

9852

9853 - `type: "grep"`

9854

9855 - `"grep"`

9856

9857 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

9858

9859 Configuration for the web_fetch tool.

9860

9861 - `enabled: boolean`

9862

9863 - `name: "web_fetch"`

9864

9865 - `"web_fetch"`

9866

9867 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

9868

9869 Permission policy for tool execution.

9870

9871 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

9872

9873 Tool calls are automatically approved without user confirmation.

9874

9875 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

9876

9877 Tool calls require user confirmation before execution.

9878

9879 - `type: "web_fetch"`

9880

9881 - `"web_fetch"`

9882

9883 - `allowed_domains: optional array of string`

9884

9885 - `blocked_domains: optional array of string`

9886

9887 - `max_content_tokens: optional number or null`

9888

9889 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

9890

9891 Configuration for the web_search tool.

9892

9893 - `enabled: boolean`

9894

9895 - `name: "web_search"`

9896

9897 - `"web_search"`

9898

9899 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

9900

9901 Permission policy for tool execution.

9902

9903 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

9904

9905 Tool calls are automatically approved without user confirmation.

9906

9907 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

9908

9909 Tool calls require user confirmation before execution.

9910

9911 - `type: "web_search"`

9912

9913 - `"web_search"`

9914

9915 - `allowed_domains: optional array of string`

9916

9917 - `blocked_domains: optional array of string`

9918

9919 - `user_location: optional BetaManagedAgentsUserLocation or null`

9920

9921 Approximate user location for search result localization.

9922

9923 - `type: "approximate"`

9924

9925 Location precision. Only "approximate" is supported.

9926

9927 - `"approximate"`

9928

9929 - `city: optional string or null`

9930

9931 City name.

9932

9933 - `country: optional string or null`

9934

9935 Two-letter ISO 3166-1 country code, uppercase.

9936

9937 - `region: optional string or null`

9938

9939 Region or state name.

9940

9941 - `timezone: optional string or null`

9942

9943 IANA timezone identifier, e.g. "America/Los_Angeles".

61219944

61229945 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

61239946

from line 10117

629410117 "name": "bash",

629510118 "permission_policy": {

629610119 "type": "always_allow"

6297 }

10120 },

10121 "type": "bash"

629810122 }

629910123 ],

630010124 "default_config": {

630110125

api/beta/agents/archive Changed · +253 / -42 lines

from line 319

319319

320320 - `configs: array of BetaManagedAgentsAgentToolConfig`

321321

322 - `enabled: boolean`

322 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

323323

324 - `name: "bash" or "edit" or "read" or 5 more`

324 Configuration for the bash tool.

325325

326 Built-in agent tool identifier.

326 - `enabled: boolean`

327327

328 - `"bash"`

328 - `name: "bash"`

329329

330 - `"edit"`

330 - `"bash"`

331331

332 - `"read"`

332 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

333333

334 - `"write"`

334 Permission policy for tool execution.

335335

336 - `"glob"`

336 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

337337

338 - `"grep"`

338 Tool calls are automatically approved without user confirmation.

339339

340 - `"web_fetch"`

340 - `type: "always_allow"`

341341

342 - `"web_search"`

342 - `"always_allow"`

343343

344 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

344 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

345345

346 Permission policy for tool execution.

346 Tool calls require user confirmation before execution.

347347

348 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

348 - `type: "always_ask"`

349349

350 Tool calls are automatically approved without user confirmation.

350 - `"always_ask"`

351351

352 - `type: "always_allow"`

352 - `type: "bash"`

353353

354 - `"always_allow"`

354 - `"bash"`

355355

356 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

356 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

357357

358 Tool calls require user confirmation before execution.

358 Configuration for the edit tool.

359359

360 - `type: "always_ask"`

360 - `enabled: boolean`

361361

362 - `"always_ask"`

362 - `name: "edit"`

363363

364 - `"edit"`

365

366 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

367

368 Permission policy for tool execution.

369

370 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

371

372 Tool calls are automatically approved without user confirmation.

373

374 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

375

376 Tool calls require user confirmation before execution.

377

378 - `type: "edit"`

379

380 - `"edit"`

381

382 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

383

384 Configuration for the read tool.

385

386 - `enabled: boolean`

387

388 - `name: "read"`

389

390 - `"read"`

391

392 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

393

394 Permission policy for tool execution.

395

396 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

397

398 Tool calls are automatically approved without user confirmation.

399

400 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

401

402 Tool calls require user confirmation before execution.

403

404 - `type: "read"`

405

406 - `"read"`

407

408 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

409

410 Configuration for the write tool.

411

412 - `enabled: boolean`

413

414 - `name: "write"`

415

416 - `"write"`

417

418 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

419

420 Permission policy for tool execution.

421

422 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

423

424 Tool calls are automatically approved without user confirmation.

425

426 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

427

428 Tool calls require user confirmation before execution.

429

430 - `type: "write"`

431

432 - `"write"`

433

434 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

435

436 Configuration for the glob tool.

437

438 - `enabled: boolean`

439

440 - `name: "glob"`

441

442 - `"glob"`

443

444 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

445

446 Permission policy for tool execution.

447

448 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

449

450 Tool calls are automatically approved without user confirmation.

451

452 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

453

454 Tool calls require user confirmation before execution.

455

456 - `type: "glob"`

457

458 - `"glob"`

459

460 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

461

462 Configuration for the grep tool.

463

464 - `enabled: boolean`

465

466 - `name: "grep"`

467

468 - `"grep"`

469

470 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

471

472 Permission policy for tool execution.

473

474 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

475

476 Tool calls are automatically approved without user confirmation.

477

478 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

479

480 Tool calls require user confirmation before execution.

481

482 - `type: "grep"`

483

484 - `"grep"`

485

486 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

487

488 Configuration for the web_fetch tool.

489

490 - `enabled: boolean`

491

492 - `name: "web_fetch"`

493

494 - `"web_fetch"`

495

496 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

497

498 Permission policy for tool execution.

499

500 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

501

502 Tool calls are automatically approved without user confirmation.

503

504 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

505

506 Tool calls require user confirmation before execution.

507

508 - `type: "web_fetch"`

509

510 - `"web_fetch"`

511

512 - `allowed_domains: optional array of string`

513

514 - `blocked_domains: optional array of string`

515

516 - `max_content_tokens: optional number or null`

517

518 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

519

520 Configuration for the web_search tool.

521

522 - `enabled: boolean`

523

524 - `name: "web_search"`

525

526 - `"web_search"`

527

528 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

529

530 Permission policy for tool execution.

531

532 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

533

534 Tool calls are automatically approved without user confirmation.

535

536 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

537

538 Tool calls require user confirmation before execution.

539

540 - `type: "web_search"`

541

542 - `"web_search"`

543

544 - `allowed_domains: optional array of string`

545

546 - `blocked_domains: optional array of string`

547

548 - `user_location: optional BetaManagedAgentsUserLocation or null`

549

550 Approximate user location for search result localization.

551

552 - `type: "approximate"`

553

554 Location precision. Only "approximate" is supported.

555

556 - `"approximate"`

557

558 - `city: optional string or null`

559

560 City name.

561

562 - `country: optional string or null`

563

564 Two-letter ISO 3166-1 country code, uppercase.

565

566 - `region: optional string or null`

567

568 Region or state name.

569

570 - `timezone: optional string or null`

571

572 IANA timezone identifier, e.g. "America/Los_Angeles".

573

364574 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

365575

366576 Resolved default configuration for agent tools.

from line 741

531741 "name": "bash",

532742 "permission_policy": {

533743 "type": "always_allow"

534 }

744 },

745 "type": "bash"

535746 }

536747 ],

537748 "default_config": {

api/beta/agents/create Changed · +575 / -114 lines

The two sides of this change are more than 400 edits apart, too far apart to line up, so this is the differ's own diff of it and the words inside a line are not marked.

from line 373

373373

374374 Per-tool configuration overrides.

375375

376 - `name: "bash" or "edit" or "read" or 5 more`

377

378 Built-in agent tool identifier.

379

380 - `"bash"`

381

382 - `"edit"`

383

384 - `"read"`

385

386 - `"write"`

387

388 - `"glob"`

389

390 - `"grep"`

391

392 - `"web_fetch"`

393

394 - `"web_search"`

376 - `BetaManagedAgentsBashToolConfigParams object { name, enabled, permission_policy, type }`

377

378 Configuration override for the bash tool.

379

380 - `name: "bash"`

381

382 Must be "bash".

383

384 - `"bash"`

385

386 - `enabled: optional boolean or null`

387

388 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

389

390 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

391

392 Permission policy for tool execution.

393

394 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

395

396 Tool calls are automatically approved without user confirmation.

397

398 - `type: "always_allow"`

399

400 - `"always_allow"`

401

402 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

403

404 Tool calls require user confirmation before execution.

405

406 - `type: "always_ask"`

407

408 - `"always_ask"`

409

410 - `type: optional "bash"`

411

412 - `"bash"`

413

414 - `BetaManagedAgentsEditToolConfigParams object { name, enabled, permission_policy, type }`

415

416 Configuration override for the edit tool.

417

418 - `name: "edit"`

419

420 Must be "edit".

421

422 - `"edit"`

423

424 - `enabled: optional boolean or null`

425

426 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

427

428 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

429

430 Permission policy for tool execution.

431

432 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

433

434 Tool calls are automatically approved without user confirmation.

435

436 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

437

438 Tool calls require user confirmation before execution.

439

440 - `type: optional "edit"`

441

442 - `"edit"`

443

444 - `BetaManagedAgentsReadToolConfigParams object { name, enabled, permission_policy, type }`

445

446 Configuration override for the read tool.

447

448 - `name: "read"`

449

450 Must be "read".

451

452 - `"read"`

453

454 - `enabled: optional boolean or null`

455

456 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

457

458 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

459

460 Permission policy for tool execution.

461

462 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

463

464 Tool calls are automatically approved without user confirmation.

465

466 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

467

468 Tool calls require user confirmation before execution.

469

470 - `type: optional "read"`

471

472 - `"read"`

473

474 - `BetaManagedAgentsWriteToolConfigParams object { name, enabled, permission_policy, type }`

475

476 Configuration override for the write tool.

477

478 - `name: "write"`

479

480 Must be "write".

481

482 - `"write"`

483

484 - `enabled: optional boolean or null`

485

486 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

487

488 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

489

490 Permission policy for tool execution.

491

492 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

493

494 Tool calls are automatically approved without user confirmation.

495

496 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

497

498 Tool calls require user confirmation before execution.

499

500 - `type: optional "write"`

501

502 - `"write"`

503

504 - `BetaManagedAgentsGlobToolConfigParams object { name, enabled, permission_policy, type }`

505

506 Configuration override for the glob tool.

507

508 - `name: "glob"`

509

510 Must be "glob".

511

512 - `"glob"`

513

514 - `enabled: optional boolean or null`

515

516 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

517

518 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

519

520 Permission policy for tool execution.

521

522 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

523

524 Tool calls are automatically approved without user confirmation.

525

526 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

527

528 Tool calls require user confirmation before execution.

529

530 - `type: optional "glob"`

531

532 - `"glob"`

533

534 - `BetaManagedAgentsGrepToolConfigParams object { name, enabled, permission_policy, type }`

535

536 Configuration override for the grep tool.

537

538 - `name: "grep"`

539

540 Must be "grep".

541

542 - `"grep"`

543

544 - `enabled: optional boolean or null`

545

546 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

547

548 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

549

550 Permission policy for tool execution.

551

552 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

553

554 Tool calls are automatically approved without user confirmation.

555

556 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

557

558 Tool calls require user confirmation before execution.

559

560 - `type: optional "grep"`

561

562 - `"grep"`

563

564 - `BetaManagedAgentsWebFetchToolConfigParams object { name, allowed_domains, blocked_domains, 4 more }`

565

566 Configuration override for the web_fetch tool.

567

568 - `name: "web_fetch"`

569

570 Must be "web_fetch".

571

572 - `"web_fetch"`

573

574 - `allowed_domains: optional array of string`

575

576 Only fetch URLs whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "docs.example.com" (no scheme, port, or path). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with blocked_domains.

577

578 - `blocked_domains: optional array of string`

579

580 Never fetch URLs whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "ads.example.com" (no scheme, port, or path). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with allowed_domains.

581

582 - `enabled: optional boolean or null`

583

584 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

585

586 - `max_content_tokens: optional number or null`

587

588 Maximum number of tokens of fetched text content to include in context per call. Does not apply to binary content such as PDFs.

589

590 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

591

592 Permission policy for tool execution.

593

594 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

595

596 Tool calls are automatically approved without user confirmation.

597

598 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

599

600 Tool calls require user confirmation before execution.

601

602 - `type: optional "web_fetch"`

603

604 - `"web_fetch"`

605

606 - `BetaManagedAgentsWebSearchToolConfigParams object { name, allowed_domains, blocked_domains, 4 more }`

607

608 Configuration override for the web_search tool.

609

610 - `name: "web_search"`

611

612 Must be "web_search".

613

614 - `"web_search"`

615

616 - `allowed_domains: optional array of string`

617

618 Only return search results whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "docs.example.com" (no scheme or port; an optional path suffix is accepted). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with blocked_domains.

619

620 - `blocked_domains: optional array of string`

621

622 Never return search results whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "ads.example.com" (no scheme or port; an optional path suffix is accepted). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with allowed_domains.

623

624 - `enabled: optional boolean or null`

625

626 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

627

628 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

629

630 Permission policy for tool execution.

631

632 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

633

634 Tool calls are automatically approved without user confirmation.

635

636 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

637

638 Tool calls require user confirmation before execution.

639

640 - `type: optional "web_search"`

641

642 - `"web_search"`

643

644 - `user_location: optional BetaManagedAgentsUserLocation or null`

645

646 Approximate user location for search result localization.

647

648 - `type: "approximate"`

649

650 Location precision. Only "approximate" is supported.

651

652 - `"approximate"`

653

654 - `city: optional string or null`

655

656 City name.

657

658 - `country: optional string or null`

659

660 Two-letter ISO 3166-1 country code, uppercase.

661

662 - `region: optional string or null`

663

664 Region or state name.

665

666 - `timezone: optional string or null`

667

668 IANA timezone identifier, e.g. "America/Los_Angeles".

669

670 - `default_config: optional BetaManagedAgentsAgentToolsetDefaultConfigParams or null`

671

672 Default configuration for all tools in a toolset.

395673

396674 - `enabled: optional boolean or null`

397675

398 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

676 Whether tools are enabled and available to Claude by default. Defaults to true if not specified.

399677

400678 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

401679

from line 683

405683

406684 Tool calls are automatically approved without user confirmation.

407685

408 - `type: "always_allow"`

409

410 - `"always_allow"`

411

412686 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

413687

414688 Tool calls require user confirmation before execution.

415689

416 - `type: "always_ask"`

417

418 - `"always_ask"`

419

420 - `default_config: optional BetaManagedAgentsAgentToolsetDefaultConfigParams or null`

421

422 Default configuration for all tools in a toolset.

690 - `BetaManagedAgentsMCPToolsetParams object { mcp_server_name, type, configs, default_config }`

691

692 Configuration for tools from an MCP server defined in `mcp_servers`.

693

694 - `mcp_server_name: string`

695

696 Name of the MCP server. Must match a server name from the mcp_servers array. 1-255 characters.

697

698 - `type: "mcp_toolset"`

699

700 - `"mcp_toolset"`

701

702 - `configs: optional array of BetaManagedAgentsMCPToolConfigParams`

703

704 Per-tool configuration overrides.

705

706 - `name: string`

707

708 Name of the MCP tool to configure. 1-128 characters.

423709

424710 - `enabled: optional boolean or null`

425711

426 Whether tools are enabled and available to Claude by default. Defaults to true if not specified.

712 Whether this tool is enabled. Overrides the `default_config` setting.

427713

428714 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

429715

from line 723

437723

438724 Tool calls require user confirmation before execution.

439725

440 - `BetaManagedAgentsMCPToolsetParams object { mcp_server_name, type, configs, default_config }`

441

442 Configuration for tools from an MCP server defined in `mcp_servers`.

443

444 - `mcp_server_name: string`

445

446 Name of the MCP server. Must match a server name from the mcp_servers array. 1-255 characters.

447

448 - `type: "mcp_toolset"`

449

450 - `"mcp_toolset"`

451

452 - `configs: optional array of BetaManagedAgentsMCPToolConfigParams`

453

454 Per-tool configuration overrides.

455

456 - `name: string`

457

458 Name of the MCP tool to configure. 1-128 characters.

726 - `default_config: optional BetaManagedAgentsMCPToolsetDefaultConfigParams or null`

727

728 Default configuration for all tools from an MCP server.

459729

460730 - `enabled: optional boolean or null`

461731

462 Whether this tool is enabled. Overrides the `default_config` setting.

732 Whether tools are enabled by default. Defaults to true if not specified.

463733

464734 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

465735

from line 743

473743

474744 Tool calls require user confirmation before execution.

475745

476 - `default_config: optional BetaManagedAgentsMCPToolsetDefaultConfigParams or null`

477

478 Default configuration for all tools from an MCP server.

479

480 - `enabled: optional boolean or null`

481

482 Whether tools are enabled by default. Defaults to true if not specified.

483

484 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

485

486 Permission policy for tool execution.

487

488 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

489

490 Tool calls are automatically approved without user confirmation.

491

492 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

493

494 Tool calls require user confirmation before execution.

495

496746 - `BetaManagedAgentsCustomToolParams object { description, input_schema, name, type }`

497747

498748 A custom tool that is executed by the API client rather than the agent. When the agent calls this tool, an `agent.custom_tool_use` event is emitted and the session goes idle, waiting for the client to provide the result via a `user.custom_tool_result` event.

from line 999

749999

7501000 - `configs: array of BetaManagedAgentsAgentToolConfig`

7511001

752 - `enabled: boolean`

753

754 - `name: "bash" or "edit" or "read" or 5 more`

755

756 Built-in agent tool identifier.

757

758 - `"bash"`

759

760 - `"edit"`

761

762 - `"read"`

763

764 - `"write"`

765

766 - `"glob"`

767

768 - `"grep"`

769

770 - `"web_fetch"`

771

772 - `"web_search"`

773

774 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

775

776 Permission policy for tool execution.

777

778 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

779

780 Tool calls are automatically approved without user confirmation.

781

782 - `type: "always_allow"`

783

784 - `"always_allow"`

785

786 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

787

788 Tool calls require user confirmation before execution.

789

790 - `type: "always_ask"`

791

792 - `"always_ask"`

1002 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

1003

1004 Configuration for the bash tool.

1005

1006 - `enabled: boolean`

1007

1008 - `name: "bash"`

1009

1010 - `"bash"`

1011

1012 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1013

1014 Permission policy for tool execution.

1015

1016 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1017

1018 Tool calls are automatically approved without user confirmation.

1019

1020 - `type: "always_allow"`

1021

1022 - `"always_allow"`

1023

1024 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1025

1026 Tool calls require user confirmation before execution.

1027

1028 - `type: "always_ask"`

1029

1030 - `"always_ask"`

1031

1032 - `type: "bash"`

1033

1034 - `"bash"`

1035

1036 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

1037

1038 Configuration for the edit tool.

1039

1040 - `enabled: boolean`

1041

1042 - `name: "edit"`

1043

1044 - `"edit"`

1045

1046 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1047

1048 Permission policy for tool execution.

1049

1050 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1051

1052 Tool calls are automatically approved without user confirmation.

1053

1054 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1055

1056 Tool calls require user confirmation before execution.

1057

1058 - `type: "edit"`

1059

1060 - `"edit"`

1061

1062 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

1063

1064 Configuration for the read tool.

1065

1066 - `enabled: boolean`

1067

1068 - `name: "read"`

1069

1070 - `"read"`

1071

1072 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1073

1074 Permission policy for tool execution.

1075

1076 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1077

1078 Tool calls are automatically approved without user confirmation.

1079

1080 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1081

1082 Tool calls require user confirmation before execution.

1083

1084 - `type: "read"`

1085

1086 - `"read"`

1087

1088 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

1089

1090 Configuration for the write tool.

1091

1092 - `enabled: boolean`

1093

1094 - `name: "write"`

1095

1096 - `"write"`

1097

1098 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1099

1100 Permission policy for tool execution.

1101

1102 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1103

1104 Tool calls are automatically approved without user confirmation.

1105

1106 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1107

1108 Tool calls require user confirmation before execution.

1109

1110 - `type: "write"`

1111

1112 - `"write"`

1113

1114 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

1115

1116 Configuration for the glob tool.

1117

1118 - `enabled: boolean`

1119

1120 - `name: "glob"`

1121

1122 - `"glob"`

1123

1124 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1125

1126 Permission policy for tool execution.

1127

1128 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1129

1130 Tool calls are automatically approved without user confirmation.

1131

1132 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1133

1134 Tool calls require user confirmation before execution.

1135

1136 - `type: "glob"`

1137

1138 - `"glob"`

1139

1140 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

1141

1142 Configuration for the grep tool.

1143

1144 - `enabled: boolean`

1145

1146 - `name: "grep"`

1147

1148 - `"grep"`

1149

1150 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1151

1152 Permission policy for tool execution.

1153

1154 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1155

1156 Tool calls are automatically approved without user confirmation.

1157

1158 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1159

1160 Tool calls require user confirmation before execution.

1161

1162 - `type: "grep"`

1163

1164 - `"grep"`

1165

1166 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

1167

1168 Configuration for the web_fetch tool.

1169

1170 - `enabled: boolean`

1171

1172 - `name: "web_fetch"`

1173

1174 - `"web_fetch"`

1175

1176 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1177

1178 Permission policy for tool execution.

1179

1180 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1181

1182 Tool calls are automatically approved without user confirmation.

1183

1184 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1185

1186 Tool calls require user confirmation before execution.

1187

1188 - `type: "web_fetch"`

1189

1190 - `"web_fetch"`

1191

1192 - `allowed_domains: optional array of string`

1193

1194 - `blocked_domains: optional array of string`

1195

1196 - `max_content_tokens: optional number or null`

1197

1198 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

1199

1200 Configuration for the web_search tool.

1201

1202 - `enabled: boolean`

1203

1204 - `name: "web_search"`

1205

1206 - `"web_search"`

1207

1208 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1209

1210 Permission policy for tool execution.

1211

1212 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1213

1214 Tool calls are automatically approved without user confirmation.

1215

1216 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1217

1218 Tool calls require user confirmation before execution.

1219

1220 - `type: "web_search"`

1221

1222 - `"web_search"`

1223

1224 - `allowed_domains: optional array of string`

1225

1226 - `blocked_domains: optional array of string`

1227

1228 - `user_location: optional BetaManagedAgentsUserLocation or null`

1229

1230 Approximate user location for search result localization.

1231

1232 - `type: "approximate"`

1233

1234 Location precision. Only "approximate" is supported.

1235

1236 - `"approximate"`

1237

1238 - `city: optional string or null`

1239

1240 City name.

1241

1242 - `country: optional string or null`

1243

1244 Two-letter ISO 3166-1 country code, uppercase.

1245

1246 - `region: optional string or null`

1247

1248 Region or state name.

1249

1250 - `timezone: optional string or null`

1251

1252 IANA timezone identifier, e.g. "America/Los_Angeles".

7931253

7941254 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

7951255

from line 1435

9751435 "name": "bash",

9761436 "permission_policy": {

9771437 "type": "always_allow"

978 }

1438 },

1439 "type": "bash"

9791440 }

9801441 ],

9811442 "default_config": {

9821443

api/beta/agents/list Changed · +253 / -42 lines

from line 337

337337

338338 - `configs: array of BetaManagedAgentsAgentToolConfig`

339339

340 - `enabled: boolean`

340 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

341341

342 - `name: "bash" or "edit" or "read" or 5 more`

342 Configuration for the bash tool.

343343

344 Built-in agent tool identifier.

344 - `enabled: boolean`

345345

346 - `"bash"`

346 - `name: "bash"`

347347

348 - `"edit"`

348 - `"bash"`

349349

350 - `"read"`

350 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

351351

352 - `"write"`

352 Permission policy for tool execution.

353353

354 - `"glob"`

354 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

355355

356 - `"grep"`

356 Tool calls are automatically approved without user confirmation.

357357

358 - `"web_fetch"`

358 - `type: "always_allow"`

359359

360 - `"web_search"`

360 - `"always_allow"`

361361

362 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

362 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

363363

364 Permission policy for tool execution.

364 Tool calls require user confirmation before execution.

365365

366 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

366 - `type: "always_ask"`

367367

368 Tool calls are automatically approved without user confirmation.

368 - `"always_ask"`

369369

370 - `type: "always_allow"`

370 - `type: "bash"`

371371

372 - `"always_allow"`

372 - `"bash"`

373373

374 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

374 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

375375

376 Tool calls require user confirmation before execution.

376 Configuration for the edit tool.

377377

378 - `type: "always_ask"`

378 - `enabled: boolean`

379379

380 - `"always_ask"`

380 - `name: "edit"`

381381

382 - `"edit"`

383

384 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

385

386 Permission policy for tool execution.

387

388 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

389

390 Tool calls are automatically approved without user confirmation.

391

392 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

393

394 Tool calls require user confirmation before execution.

395

396 - `type: "edit"`

397

398 - `"edit"`

399

400 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

401

402 Configuration for the read tool.

403

404 - `enabled: boolean`

405

406 - `name: "read"`

407

408 - `"read"`

409

410 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

411

412 Permission policy for tool execution.

413

414 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

415

416 Tool calls are automatically approved without user confirmation.

417

418 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

419

420 Tool calls require user confirmation before execution.

421

422 - `type: "read"`

423

424 - `"read"`

425

426 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

427

428 Configuration for the write tool.

429

430 - `enabled: boolean`

431

432 - `name: "write"`

433

434 - `"write"`

435

436 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

437

438 Permission policy for tool execution.

439

440 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

441

442 Tool calls are automatically approved without user confirmation.

443

444 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

445

446 Tool calls require user confirmation before execution.

447

448 - `type: "write"`

449

450 - `"write"`

451

452 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

453

454 Configuration for the glob tool.

455

456 - `enabled: boolean`

457

458 - `name: "glob"`

459

460 - `"glob"`

461

462 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

463

464 Permission policy for tool execution.

465

466 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

467

468 Tool calls are automatically approved without user confirmation.

469

470 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

471

472 Tool calls require user confirmation before execution.

473

474 - `type: "glob"`

475

476 - `"glob"`

477

478 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

479

480 Configuration for the grep tool.

481

482 - `enabled: boolean`

483

484 - `name: "grep"`

485

486 - `"grep"`

487

488 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

489

490 Permission policy for tool execution.

491

492 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

493

494 Tool calls are automatically approved without user confirmation.

495

496 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

497

498 Tool calls require user confirmation before execution.

499

500 - `type: "grep"`

501

502 - `"grep"`

503

504 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

505

506 Configuration for the web_fetch tool.

507

508 - `enabled: boolean`

509

510 - `name: "web_fetch"`

511

512 - `"web_fetch"`

513

514 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

515

516 Permission policy for tool execution.

517

518 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

519

520 Tool calls are automatically approved without user confirmation.

521

522 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

523

524 Tool calls require user confirmation before execution.

525

526 - `type: "web_fetch"`

527

528 - `"web_fetch"`

529

530 - `allowed_domains: optional array of string`

531

532 - `blocked_domains: optional array of string`

533

534 - `max_content_tokens: optional number or null`

535

536 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

537

538 Configuration for the web_search tool.

539

540 - `enabled: boolean`

541

542 - `name: "web_search"`

543

544 - `"web_search"`

545

546 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

547

548 Permission policy for tool execution.

549

550 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

551

552 Tool calls are automatically approved without user confirmation.

553

554 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

555

556 Tool calls require user confirmation before execution.

557

558 - `type: "web_search"`

559

560 - `"web_search"`

561

562 - `allowed_domains: optional array of string`

563

564 - `blocked_domains: optional array of string`

565

566 - `user_location: optional BetaManagedAgentsUserLocation or null`

567

568 Approximate user location for search result localization.

569

570 - `type: "approximate"`

571

572 Location precision. Only "approximate" is supported.

573

574 - `"approximate"`

575

576 - `city: optional string or null`

577

578 City name.

579

580 - `country: optional string or null`

581

582 Two-letter ISO 3166-1 country code, uppercase.

583

584 - `region: optional string or null`

585

586 Region or state name.

587

588 - `timezone: optional string or null`

589

590 IANA timezone identifier, e.g. "America/Los_Angeles".

591

382592 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

383593

384594 Resolved default configuration for agent tools.

from line 764

554764 "name": "bash",

555765 "permission_policy": {

556766 "type": "always_allow"

557 }

767 },

768 "type": "bash"

558769 }

559770 ],

560771 "default_config": {

api/beta/agents/retrieve Changed · +253 / -42 lines

from line 325

325325

326326 - `configs: array of BetaManagedAgentsAgentToolConfig`

327327

328 - `enabled: boolean`

328 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

329329

330 - `name: "bash" or "edit" or "read" or 5 more`

330 Configuration for the bash tool.

331331

332 Built-in agent tool identifier.

332 - `enabled: boolean`

333333

334 - `"bash"`

334 - `name: "bash"`

335335

336 - `"edit"`

336 - `"bash"`

337337

338 - `"read"`

338 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

339339

340 - `"write"`

340 Permission policy for tool execution.

341341

342 - `"glob"`

342 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

343343

344 - `"grep"`

344 Tool calls are automatically approved without user confirmation.

345345

346 - `"web_fetch"`

346 - `type: "always_allow"`

347347

348 - `"web_search"`

348 - `"always_allow"`

349349

350 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

350 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

351351

352 Permission policy for tool execution.

352 Tool calls require user confirmation before execution.

353353

354 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

354 - `type: "always_ask"`

355355

356 Tool calls are automatically approved without user confirmation.

356 - `"always_ask"`

357357

358 - `type: "always_allow"`

358 - `type: "bash"`

359359

360 - `"always_allow"`

360 - `"bash"`

361361

362 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

362 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

363363

364 Tool calls require user confirmation before execution.

364 Configuration for the edit tool.

365365

366 - `type: "always_ask"`

366 - `enabled: boolean`

367367

368 - `"always_ask"`

368 - `name: "edit"`

369369

370 - `"edit"`

371

372 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

373

374 Permission policy for tool execution.

375

376 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

377

378 Tool calls are automatically approved without user confirmation.

379

380 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

381

382 Tool calls require user confirmation before execution.

383

384 - `type: "edit"`

385

386 - `"edit"`

387

388 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

389

390 Configuration for the read tool.

391

392 - `enabled: boolean`

393

394 - `name: "read"`

395

396 - `"read"`

397

398 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

399

400 Permission policy for tool execution.

401

402 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

403

404 Tool calls are automatically approved without user confirmation.

405

406 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

407

408 Tool calls require user confirmation before execution.

409

410 - `type: "read"`

411

412 - `"read"`

413

414 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

415

416 Configuration for the write tool.

417

418 - `enabled: boolean`

419

420 - `name: "write"`

421

422 - `"write"`

423

424 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

425

426 Permission policy for tool execution.

427

428 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

429

430 Tool calls are automatically approved without user confirmation.

431

432 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

433

434 Tool calls require user confirmation before execution.

435

436 - `type: "write"`

437

438 - `"write"`

439

440 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

441

442 Configuration for the glob tool.

443

444 - `enabled: boolean`

445

446 - `name: "glob"`

447

448 - `"glob"`

449

450 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

451

452 Permission policy for tool execution.

453

454 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

455

456 Tool calls are automatically approved without user confirmation.

457

458 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

459

460 Tool calls require user confirmation before execution.

461

462 - `type: "glob"`

463

464 - `"glob"`

465

466 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

467

468 Configuration for the grep tool.

469

470 - `enabled: boolean`

471

472 - `name: "grep"`

473

474 - `"grep"`

475

476 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

477

478 Permission policy for tool execution.

479

480 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

481

482 Tool calls are automatically approved without user confirmation.

483

484 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

485

486 Tool calls require user confirmation before execution.

487

488 - `type: "grep"`

489

490 - `"grep"`

491

492 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

493

494 Configuration for the web_fetch tool.

495

496 - `enabled: boolean`

497

498 - `name: "web_fetch"`

499

500 - `"web_fetch"`

501

502 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

503

504 Permission policy for tool execution.

505

506 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

507

508 Tool calls are automatically approved without user confirmation.

509

510 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

511

512 Tool calls require user confirmation before execution.

513

514 - `type: "web_fetch"`

515

516 - `"web_fetch"`

517

518 - `allowed_domains: optional array of string`

519

520 - `blocked_domains: optional array of string`

521

522 - `max_content_tokens: optional number or null`

523

524 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

525

526 Configuration for the web_search tool.

527

528 - `enabled: boolean`

529

530 - `name: "web_search"`

531

532 - `"web_search"`

533

534 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

535

536 Permission policy for tool execution.

537

538 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

539

540 Tool calls are automatically approved without user confirmation.

541

542 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

543

544 Tool calls require user confirmation before execution.

545

546 - `type: "web_search"`

547

548 - `"web_search"`

549

550 - `allowed_domains: optional array of string`

551

552 - `blocked_domains: optional array of string`

553

554 - `user_location: optional BetaManagedAgentsUserLocation or null`

555

556 Approximate user location for search result localization.

557

558 - `type: "approximate"`

559

560 Location precision. Only "approximate" is supported.

561

562 - `"approximate"`

563

564 - `city: optional string or null`

565

566 City name.

567

568 - `country: optional string or null`

569

570 Two-letter ISO 3166-1 country code, uppercase.

571

572 - `region: optional string or null`

573

574 Region or state name.

575

576 - `timezone: optional string or null`

577

578 IANA timezone identifier, e.g. "America/Los_Angeles".

579

370580 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

371581

372582 Resolved default configuration for agent tools.

from line 746

536746 "name": "bash",

537747 "permission_policy": {

538748 "type": "always_allow"

539 }

749 },

750 "type": "bash"

540751 }

541752 ],

542753 "default_config": {

api/beta/agents/update Changed · +575 / -114 lines

The two sides of this change are more than 400 edits apart, too far apart to line up, so this is the differ's own diff of it and the words inside a line are not marked.

from line 377

377377

378378 Per-tool configuration overrides.

379379

380 - `name: "bash" or "edit" or "read" or 5 more`

381

382 Built-in agent tool identifier.

383

384 - `"bash"`

385

386 - `"edit"`

387

388 - `"read"`

389

390 - `"write"`

391

392 - `"glob"`

393

394 - `"grep"`

395

396 - `"web_fetch"`

397

398 - `"web_search"`

380 - `BetaManagedAgentsBashToolConfigParams object { name, enabled, permission_policy, type }`

381

382 Configuration override for the bash tool.

383

384 - `name: "bash"`

385

386 Must be "bash".

387

388 - `"bash"`

389

390 - `enabled: optional boolean or null`

391

392 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

393

394 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

395

396 Permission policy for tool execution.

397

398 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

399

400 Tool calls are automatically approved without user confirmation.

401

402 - `type: "always_allow"`

403

404 - `"always_allow"`

405

406 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

407

408 Tool calls require user confirmation before execution.

409

410 - `type: "always_ask"`

411

412 - `"always_ask"`

413

414 - `type: optional "bash"`

415

416 - `"bash"`

417

418 - `BetaManagedAgentsEditToolConfigParams object { name, enabled, permission_policy, type }`

419

420 Configuration override for the edit tool.

421

422 - `name: "edit"`

423

424 Must be "edit".

425

426 - `"edit"`

427

428 - `enabled: optional boolean or null`

429

430 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

431

432 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

433

434 Permission policy for tool execution.

435

436 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

437

438 Tool calls are automatically approved without user confirmation.

439

440 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

441

442 Tool calls require user confirmation before execution.

443

444 - `type: optional "edit"`

445

446 - `"edit"`

447

448 - `BetaManagedAgentsReadToolConfigParams object { name, enabled, permission_policy, type }`

449

450 Configuration override for the read tool.

451

452 - `name: "read"`

453

454 Must be "read".

455

456 - `"read"`

457

458 - `enabled: optional boolean or null`

459

460 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

461

462 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

463

464 Permission policy for tool execution.

465

466 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

467

468 Tool calls are automatically approved without user confirmation.

469

470 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

471

472 Tool calls require user confirmation before execution.

473

474 - `type: optional "read"`

475

476 - `"read"`

477

478 - `BetaManagedAgentsWriteToolConfigParams object { name, enabled, permission_policy, type }`

479

480 Configuration override for the write tool.

481

482 - `name: "write"`

483

484 Must be "write".

485

486 - `"write"`

487

488 - `enabled: optional boolean or null`

489

490 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

491

492 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

493

494 Permission policy for tool execution.

495

496 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

497

498 Tool calls are automatically approved without user confirmation.

499

500 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

501

502 Tool calls require user confirmation before execution.

503

504 - `type: optional "write"`

505

506 - `"write"`

507

508 - `BetaManagedAgentsGlobToolConfigParams object { name, enabled, permission_policy, type }`

509

510 Configuration override for the glob tool.

511

512 - `name: "glob"`

513

514 Must be "glob".

515

516 - `"glob"`

517

518 - `enabled: optional boolean or null`

519

520 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

521

522 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

523

524 Permission policy for tool execution.

525

526 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

527

528 Tool calls are automatically approved without user confirmation.

529

530 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

531

532 Tool calls require user confirmation before execution.

533

534 - `type: optional "glob"`

535

536 - `"glob"`

537

538 - `BetaManagedAgentsGrepToolConfigParams object { name, enabled, permission_policy, type }`

539

540 Configuration override for the grep tool.

541

542 - `name: "grep"`

543

544 Must be "grep".

545

546 - `"grep"`

547

548 - `enabled: optional boolean or null`

549

550 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

551

552 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

553

554 Permission policy for tool execution.

555

556 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

557

558 Tool calls are automatically approved without user confirmation.

559

560 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

561

562 Tool calls require user confirmation before execution.

563

564 - `type: optional "grep"`

565

566 - `"grep"`

567

568 - `BetaManagedAgentsWebFetchToolConfigParams object { name, allowed_domains, blocked_domains, 4 more }`

569

570 Configuration override for the web_fetch tool.

571

572 - `name: "web_fetch"`

573

574 Must be "web_fetch".

575

576 - `"web_fetch"`

577

578 - `allowed_domains: optional array of string`

579

580 Only fetch URLs whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "docs.example.com" (no scheme, port, or path). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with blocked_domains.

581

582 - `blocked_domains: optional array of string`

583

584 Never fetch URLs whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "ads.example.com" (no scheme, port, or path). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with allowed_domains.

585

586 - `enabled: optional boolean or null`

587

588 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

589

590 - `max_content_tokens: optional number or null`

591

592 Maximum number of tokens of fetched text content to include in context per call. Does not apply to binary content such as PDFs.

593

594 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

595

596 Permission policy for tool execution.

597

598 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

599

600 Tool calls are automatically approved without user confirmation.

601

602 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

603

604 Tool calls require user confirmation before execution.

605

606 - `type: optional "web_fetch"`

607

608 - `"web_fetch"`

609

610 - `BetaManagedAgentsWebSearchToolConfigParams object { name, allowed_domains, blocked_domains, 4 more }`

611

612 Configuration override for the web_search tool.

613

614 - `name: "web_search"`

615

616 Must be "web_search".

617

618 - `"web_search"`

619

620 - `allowed_domains: optional array of string`

621

622 Only return search results whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "docs.example.com" (no scheme or port; an optional path suffix is accepted). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with blocked_domains.

623

624 - `blocked_domains: optional array of string`

625

626 Never return search results whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "ads.example.com" (no scheme or port; an optional path suffix is accepted). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with allowed_domains.

627

628 - `enabled: optional boolean or null`

629

630 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

631

632 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

633

634 Permission policy for tool execution.

635

636 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

637

638 Tool calls are automatically approved without user confirmation.

639

640 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

641

642 Tool calls require user confirmation before execution.

643

644 - `type: optional "web_search"`

645

646 - `"web_search"`

647

648 - `user_location: optional BetaManagedAgentsUserLocation or null`

649

650 Approximate user location for search result localization.

651

652 - `type: "approximate"`

653

654 Location precision. Only "approximate" is supported.

655

656 - `"approximate"`

657

658 - `city: optional string or null`

659

660 City name.

661

662 - `country: optional string or null`

663

664 Two-letter ISO 3166-1 country code, uppercase.

665

666 - `region: optional string or null`

667

668 Region or state name.

669

670 - `timezone: optional string or null`

671

672 IANA timezone identifier, e.g. "America/Los_Angeles".

673

674 - `default_config: optional BetaManagedAgentsAgentToolsetDefaultConfigParams or null`

675

676 Default configuration for all tools in a toolset.

399677

400678 - `enabled: optional boolean or null`

401679

402 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

680 Whether tools are enabled and available to Claude by default. Defaults to true if not specified.

403681

404682 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

405683

from line 687

409687

410688 Tool calls are automatically approved without user confirmation.

411689

412 - `type: "always_allow"`

413

414 - `"always_allow"`

415

416690 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

417691

418692 Tool calls require user confirmation before execution.

419693

420 - `type: "always_ask"`

421

422 - `"always_ask"`

423

424 - `default_config: optional BetaManagedAgentsAgentToolsetDefaultConfigParams or null`

425

426 Default configuration for all tools in a toolset.

694 - `BetaManagedAgentsMCPToolsetParams object { mcp_server_name, type, configs, default_config }`

695

696 Configuration for tools from an MCP server defined in `mcp_servers`.

697

698 - `mcp_server_name: string`

699

700 Name of the MCP server. Must match a server name from the mcp_servers array. 1-255 characters.

701

702 - `type: "mcp_toolset"`

703

704 - `"mcp_toolset"`

705

706 - `configs: optional array of BetaManagedAgentsMCPToolConfigParams`

707

708 Per-tool configuration overrides.

709

710 - `name: string`

711

712 Name of the MCP tool to configure. 1-128 characters.

427713

428714 - `enabled: optional boolean or null`

429715

430 Whether tools are enabled and available to Claude by default. Defaults to true if not specified.

716 Whether this tool is enabled. Overrides the `default_config` setting.

431717

432718 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

433719

from line 727

441727

442728 Tool calls require user confirmation before execution.

443729

444 - `BetaManagedAgentsMCPToolsetParams object { mcp_server_name, type, configs, default_config }`

445

446 Configuration for tools from an MCP server defined in `mcp_servers`.

447

448 - `mcp_server_name: string`

449

450 Name of the MCP server. Must match a server name from the mcp_servers array. 1-255 characters.

451

452 - `type: "mcp_toolset"`

453

454 - `"mcp_toolset"`

455

456 - `configs: optional array of BetaManagedAgentsMCPToolConfigParams`

457

458 Per-tool configuration overrides.

459

460 - `name: string`

461

462 Name of the MCP tool to configure. 1-128 characters.

730 - `default_config: optional BetaManagedAgentsMCPToolsetDefaultConfigParams or null`

731

732 Default configuration for all tools from an MCP server.

463733

464734 - `enabled: optional boolean or null`

465735

466 Whether this tool is enabled. Overrides the `default_config` setting.

736 Whether tools are enabled by default. Defaults to true if not specified.

467737

468738 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

469739

from line 747

477747

478748 Tool calls require user confirmation before execution.

479749

480 - `default_config: optional BetaManagedAgentsMCPToolsetDefaultConfigParams or null`

481

482 Default configuration for all tools from an MCP server.

483

484 - `enabled: optional boolean or null`

485

486 Whether tools are enabled by default. Defaults to true if not specified.

487

488 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

489

490 Permission policy for tool execution.

491

492 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

493

494 Tool calls are automatically approved without user confirmation.

495

496 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

497

498 Tool calls require user confirmation before execution.

499

500750 - `BetaManagedAgentsCustomToolParams object { description, input_schema, name, type }`

501751

502752 A custom tool that is executed by the API client rather than the agent. When the agent calls this tool, an `agent.custom_tool_use` event is emitted and the session goes idle, waiting for the client to provide the result via a `user.custom_tool_result` event.

from line 1007

7571007

7581008 - `configs: array of BetaManagedAgentsAgentToolConfig`

7591009

760 - `enabled: boolean`

761

762 - `name: "bash" or "edit" or "read" or 5 more`

763

764 Built-in agent tool identifier.

765

766 - `"bash"`

767

768 - `"edit"`

769

770 - `"read"`

771

772 - `"write"`

773

774 - `"glob"`

775

776 - `"grep"`

777

778 - `"web_fetch"`

779

780 - `"web_search"`

781

782 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

783

784 Permission policy for tool execution.

785

786 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

787

788 Tool calls are automatically approved without user confirmation.

789

790 - `type: "always_allow"`

791

792 - `"always_allow"`

793

794 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

795

796 Tool calls require user confirmation before execution.

797

798 - `type: "always_ask"`

799

800 - `"always_ask"`

1010 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

1011

1012 Configuration for the bash tool.

1013

1014 - `enabled: boolean`

1015

1016 - `name: "bash"`

1017

1018 - `"bash"`

1019

1020 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1021

1022 Permission policy for tool execution.

1023

1024 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1025

1026 Tool calls are automatically approved without user confirmation.

1027

1028 - `type: "always_allow"`

1029

1030 - `"always_allow"`

1031

1032 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1033

1034 Tool calls require user confirmation before execution.

1035

1036 - `type: "always_ask"`

1037

1038 - `"always_ask"`

1039

1040 - `type: "bash"`

1041

1042 - `"bash"`

1043

1044 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

1045

1046 Configuration for the edit tool.

1047

1048 - `enabled: boolean`

1049

1050 - `name: "edit"`

1051

1052 - `"edit"`

1053

1054 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1055

1056 Permission policy for tool execution.

1057

1058 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1059

1060 Tool calls are automatically approved without user confirmation.

1061

1062 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1063

1064 Tool calls require user confirmation before execution.

1065

1066 - `type: "edit"`

1067

1068 - `"edit"`

1069

1070 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

1071

1072 Configuration for the read tool.

1073

1074 - `enabled: boolean`

1075

1076 - `name: "read"`

1077

1078 - `"read"`

1079

1080 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1081

1082 Permission policy for tool execution.

1083

1084 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1085

1086 Tool calls are automatically approved without user confirmation.

1087

1088 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1089

1090 Tool calls require user confirmation before execution.

1091

1092 - `type: "read"`

1093

1094 - `"read"`

1095

1096 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

1097

1098 Configuration for the write tool.

1099

1100 - `enabled: boolean`

1101

1102 - `name: "write"`

1103

1104 - `"write"`

1105

1106 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1107

1108 Permission policy for tool execution.

1109

1110 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1111

1112 Tool calls are automatically approved without user confirmation.

1113

1114 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1115

1116 Tool calls require user confirmation before execution.

1117

1118 - `type: "write"`

1119

1120 - `"write"`

1121

1122 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

1123

1124 Configuration for the glob tool.

1125

1126 - `enabled: boolean`

1127

1128 - `name: "glob"`

1129

1130 - `"glob"`

1131

1132 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1133

1134 Permission policy for tool execution.

1135

1136 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1137

1138 Tool calls are automatically approved without user confirmation.

1139

1140 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1141

1142 Tool calls require user confirmation before execution.

1143

1144 - `type: "glob"`

1145

1146 - `"glob"`

1147

1148 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

1149

1150 Configuration for the grep tool.

1151

1152 - `enabled: boolean`

1153

1154 - `name: "grep"`

1155

1156 - `"grep"`

1157

1158 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1159

1160 Permission policy for tool execution.

1161

1162 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1163

1164 Tool calls are automatically approved without user confirmation.

1165

1166 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1167

1168 Tool calls require user confirmation before execution.

1169

1170 - `type: "grep"`

1171

1172 - `"grep"`

1173

1174 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

1175

1176 Configuration for the web_fetch tool.

1177

1178 - `enabled: boolean`

1179

1180 - `name: "web_fetch"`

1181

1182 - `"web_fetch"`

1183

1184 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1185

1186 Permission policy for tool execution.

1187

1188 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1189

1190 Tool calls are automatically approved without user confirmation.

1191

1192 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1193

1194 Tool calls require user confirmation before execution.

1195

1196 - `type: "web_fetch"`

1197

1198 - `"web_fetch"`

1199

1200 - `allowed_domains: optional array of string`

1201

1202 - `blocked_domains: optional array of string`

1203

1204 - `max_content_tokens: optional number or null`

1205

1206 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

1207

1208 Configuration for the web_search tool.

1209

1210 - `enabled: boolean`

1211

1212 - `name: "web_search"`

1213

1214 - `"web_search"`

1215

1216 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1217

1218 Permission policy for tool execution.

1219

1220 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1221

1222 Tool calls are automatically approved without user confirmation.

1223

1224 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1225

1226 Tool calls require user confirmation before execution.

1227

1228 - `type: "web_search"`

1229

1230 - `"web_search"`

1231

1232 - `allowed_domains: optional array of string`

1233

1234 - `blocked_domains: optional array of string`

1235

1236 - `user_location: optional BetaManagedAgentsUserLocation or null`

1237

1238 Approximate user location for search result localization.

1239

1240 - `type: "approximate"`

1241

1242 Location precision. Only "approximate" is supported.

1243

1244 - `"approximate"`

1245

1246 - `city: optional string or null`

1247

1248 City name.

1249

1250 - `country: optional string or null`

1251

1252 Two-letter ISO 3166-1 country code, uppercase.

1253

1254 - `region: optional string or null`

1255

1256 Region or state name.

1257

1258 - `timezone: optional string or null`

1259

1260 IANA timezone identifier, e.g. "America/Los_Angeles".

8011261

8021262 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

8031263

from line 1434

9741434 "name": "bash",

9751435 "permission_policy": {

9761436 "type": "always_allow"

977 }

1437 },

1438 "type": "bash"

9781439 }

9791440 ],

9801441 "default_config": {

9811442

api/beta/agents/versions Changed · +253 / -42 lines

from line 331

331331

332332 - `configs: array of BetaManagedAgentsAgentToolConfig`

333333

334 - `enabled: boolean`

334 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

335335

336 - `name: "bash" or "edit" or "read" or 5 more`

336 Configuration for the bash tool.

337337

338 Built-in agent tool identifier.

338 - `enabled: boolean`

339339

340 - `"bash"`

340 - `name: "bash"`

341341

342 - `"edit"`

342 - `"bash"`

343343

344 - `"read"`

344 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

345345

346 - `"write"`

346 Permission policy for tool execution.

347347

348 - `"glob"`

348 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

349349

350 - `"grep"`

350 Tool calls are automatically approved without user confirmation.

351351

352 - `"web_fetch"`

352 - `type: "always_allow"`

353353

354 - `"web_search"`

354 - `"always_allow"`

355355

356 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

356 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

357357

358 Permission policy for tool execution.

358 Tool calls require user confirmation before execution.

359359

360 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

360 - `type: "always_ask"`

361361

362 Tool calls are automatically approved without user confirmation.

362 - `"always_ask"`

363363

364 - `type: "always_allow"`

364 - `type: "bash"`

365365

366 - `"always_allow"`

366 - `"bash"`

367367

368 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

368 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

369369

370 Tool calls require user confirmation before execution.

370 Configuration for the edit tool.

371371

372 - `type: "always_ask"`

372 - `enabled: boolean`

373373

374 - `"always_ask"`

374 - `name: "edit"`

375375

376 - `"edit"`

377

378 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

379

380 Permission policy for tool execution.

381

382 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

383

384 Tool calls are automatically approved without user confirmation.

385

386 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

387

388 Tool calls require user confirmation before execution.

389

390 - `type: "edit"`

391

392 - `"edit"`

393

394 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

395

396 Configuration for the read tool.

397

398 - `enabled: boolean`

399

400 - `name: "read"`

401

402 - `"read"`

403

404 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

405

406 Permission policy for tool execution.

407

408 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

409

410 Tool calls are automatically approved without user confirmation.

411

412 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

413

414 Tool calls require user confirmation before execution.

415

416 - `type: "read"`

417

418 - `"read"`

419

420 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

421

422 Configuration for the write tool.

423

424 - `enabled: boolean`

425

426 - `name: "write"`

427

428 - `"write"`

429

430 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

431

432 Permission policy for tool execution.

433

434 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

435

436 Tool calls are automatically approved without user confirmation.

437

438 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

439

440 Tool calls require user confirmation before execution.

441

442 - `type: "write"`

443

444 - `"write"`

445

446 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

447

448 Configuration for the glob tool.

449

450 - `enabled: boolean`

451

452 - `name: "glob"`

453

454 - `"glob"`

455

456 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

457

458 Permission policy for tool execution.

459

460 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

461

462 Tool calls are automatically approved without user confirmation.

463

464 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

465

466 Tool calls require user confirmation before execution.

467

468 - `type: "glob"`

469

470 - `"glob"`

471

472 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

473

474 Configuration for the grep tool.

475

476 - `enabled: boolean`

477

478 - `name: "grep"`

479

480 - `"grep"`

481

482 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

483

484 Permission policy for tool execution.

485

486 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

487

488 Tool calls are automatically approved without user confirmation.

489

490 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

491

492 Tool calls require user confirmation before execution.

493

494 - `type: "grep"`

495

496 - `"grep"`

497

498 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

499

500 Configuration for the web_fetch tool.

501

502 - `enabled: boolean`

503

504 - `name: "web_fetch"`

505

506 - `"web_fetch"`

507

508 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

509

510 Permission policy for tool execution.

511

512 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

513

514 Tool calls are automatically approved without user confirmation.

515

516 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

517

518 Tool calls require user confirmation before execution.

519

520 - `type: "web_fetch"`

521

522 - `"web_fetch"`

523

524 - `allowed_domains: optional array of string`

525

526 - `blocked_domains: optional array of string`

527

528 - `max_content_tokens: optional number or null`

529

530 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

531

532 Configuration for the web_search tool.

533

534 - `enabled: boolean`

535

536 - `name: "web_search"`

537

538 - `"web_search"`

539

540 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

541

542 Permission policy for tool execution.

543

544 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

545

546 Tool calls are automatically approved without user confirmation.

547

548 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

549

550 Tool calls require user confirmation before execution.

551

552 - `type: "web_search"`

553

554 - `"web_search"`

555

556 - `allowed_domains: optional array of string`

557

558 - `blocked_domains: optional array of string`

559

560 - `user_location: optional BetaManagedAgentsUserLocation or null`

561

562 Approximate user location for search result localization.

563

564 - `type: "approximate"`

565

566 Location precision. Only "approximate" is supported.

567

568 - `"approximate"`

569

570 - `city: optional string or null`

571

572 City name.

573

574 - `country: optional string or null`

575

576 Two-letter ISO 3166-1 country code, uppercase.

577

578 - `region: optional string or null`

579

580 Region or state name.

581

582 - `timezone: optional string or null`

583

584 IANA timezone identifier, e.g. "America/Los_Angeles".

585

376586 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

377587

378588 Resolved default configuration for agent tools.

from line 758

548758 "name": "bash",

549759 "permission_policy": {

550760 "type": "always_allow"

551 }

761 },

762 "type": "bash"

552763 }

553764 ],

554765 "default_config": {

api/beta/agents/versions/list Changed · +253 / -42 lines

from line 329

329329

330330 - `configs: array of BetaManagedAgentsAgentToolConfig`

331331

332 - `enabled: boolean`

332 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

333333

334 - `name: "bash" or "edit" or "read" or 5 more`

334 Configuration for the bash tool.

335335

336 Built-in agent tool identifier.

336 - `enabled: boolean`

337337

338 - `"bash"`

338 - `name: "bash"`

339339

340 - `"edit"`

340 - `"bash"`

341341

342 - `"read"`

342 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

343343

344 - `"write"`

344 Permission policy for tool execution.

345345

346 - `"glob"`

346 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

347347

348 - `"grep"`

348 Tool calls are automatically approved without user confirmation.

349349

350 - `"web_fetch"`

350 - `type: "always_allow"`

351351

352 - `"web_search"`

352 - `"always_allow"`

353353

354 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

354 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

355355

356 Permission policy for tool execution.

356 Tool calls require user confirmation before execution.

357357

358 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

358 - `type: "always_ask"`

359359

360 Tool calls are automatically approved without user confirmation.

360 - `"always_ask"`

361361

362 - `type: "always_allow"`

362 - `type: "bash"`

363363

364 - `"always_allow"`

364 - `"bash"`

365365

366 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

366 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

367367

368 Tool calls require user confirmation before execution.

368 Configuration for the edit tool.

369369

370 - `type: "always_ask"`

370 - `enabled: boolean`

371371

372 - `"always_ask"`

372 - `name: "edit"`

373373

374 - `"edit"`

375

376 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

377

378 Permission policy for tool execution.

379

380 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

381

382 Tool calls are automatically approved without user confirmation.

383

384 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

385

386 Tool calls require user confirmation before execution.

387

388 - `type: "edit"`

389

390 - `"edit"`

391

392 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

393

394 Configuration for the read tool.

395

396 - `enabled: boolean`

397

398 - `name: "read"`

399

400 - `"read"`

401

402 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

403

404 Permission policy for tool execution.

405

406 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

407

408 Tool calls are automatically approved without user confirmation.

409

410 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

411

412 Tool calls require user confirmation before execution.

413

414 - `type: "read"`

415

416 - `"read"`

417

418 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

419

420 Configuration for the write tool.

421

422 - `enabled: boolean`

423

424 - `name: "write"`

425

426 - `"write"`

427

428 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

429

430 Permission policy for tool execution.

431

432 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

433

434 Tool calls are automatically approved without user confirmation.

435

436 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

437

438 Tool calls require user confirmation before execution.

439

440 - `type: "write"`

441

442 - `"write"`

443

444 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

445

446 Configuration for the glob tool.

447

448 - `enabled: boolean`

449

450 - `name: "glob"`

451

452 - `"glob"`

453

454 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

455

456 Permission policy for tool execution.

457

458 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

459

460 Tool calls are automatically approved without user confirmation.

461

462 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

463

464 Tool calls require user confirmation before execution.

465

466 - `type: "glob"`

467

468 - `"glob"`

469

470 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

471

472 Configuration for the grep tool.

473

474 - `enabled: boolean`

475

476 - `name: "grep"`

477

478 - `"grep"`

479

480 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

481

482 Permission policy for tool execution.

483

484 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

485

486 Tool calls are automatically approved without user confirmation.

487

488 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

489

490 Tool calls require user confirmation before execution.

491

492 - `type: "grep"`

493

494 - `"grep"`

495

496 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

497

498 Configuration for the web_fetch tool.

499

500 - `enabled: boolean`

501

502 - `name: "web_fetch"`

503

504 - `"web_fetch"`

505

506 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

507

508 Permission policy for tool execution.

509

510 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

511

512 Tool calls are automatically approved without user confirmation.

513

514 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

515

516 Tool calls require user confirmation before execution.

517

518 - `type: "web_fetch"`

519

520 - `"web_fetch"`

521

522 - `allowed_domains: optional array of string`

523

524 - `blocked_domains: optional array of string`

525

526 - `max_content_tokens: optional number or null`

527

528 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

529

530 Configuration for the web_search tool.

531

532 - `enabled: boolean`

533

534 - `name: "web_search"`

535

536 - `"web_search"`

537

538 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

539

540 Permission policy for tool execution.

541

542 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

543

544 Tool calls are automatically approved without user confirmation.

545

546 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

547

548 Tool calls require user confirmation before execution.

549

550 - `type: "web_search"`

551

552 - `"web_search"`

553

554 - `allowed_domains: optional array of string`

555

556 - `blocked_domains: optional array of string`

557

558 - `user_location: optional BetaManagedAgentsUserLocation or null`

559

560 Approximate user location for search result localization.

561

562 - `type: "approximate"`

563

564 Location precision. Only "approximate" is supported.

565

566 - `"approximate"`

567

568 - `city: optional string or null`

569

570 City name.

571

572 - `country: optional string or null`

573

574 Two-letter ISO 3166-1 country code, uppercase.

575

576 - `region: optional string or null`

577

578 Region or state name.

579

580 - `timezone: optional string or null`

581

582 IANA timezone identifier, e.g. "America/Los_Angeles".

583

374584 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

375585

376586 Resolved default configuration for agent tools.

from line 756

546756 "name": "bash",

547757 "permission_policy": {

548758 "type": "always_allow"

549 }

759 },

760 "type": "bash"

550761 }

551762 ],

552763 "default_config": {

api/beta/messages Changed · +11 / -11 lines

This page is larger than the 256 KiB this site keeps, so one side of the diff below stops where the stored text does.

from line 4000

40004000

40014001 - `version: string`

40024002

4003 Skill version or 'latest' for most recent version

4003 The resolved version: a skill version ID for custom skills.

40044004

40054005 - `content: array of BetaContentBlock`

40064006

from line 8171

81718171

81728172 - `enabled: optional boolean or null`

81738173

8174 Whether this member is offered to the model. Default is per member, per the toolset's documentation.

8174 Whether this member is offered to the model. Default is per member, per the toolset's docu

api/beta/messages/batches Changed · +4 / -4 lines

This page is larger than the 256 KiB this site keeps, so one side of the diff below stops where the stored text does.

from line 5071

50715071

50725072 - `version: string`

50735073

5074 Skill version or 'latest' for most recent version

5074 The resolved version: a skill version ID for custom skills.

50755075

50765076 - `content: array of BetaContentBlock`

50775077

from line 6962

69626962

69636963 - `version: string`

69646964

6965 Skill version or 'latest' for most recent version

6965 The resolved version: a skill version ID for custom skills.

69666966

69676967 - `content: array of BetaContentBlock`

69686968

from line 8014

80148014

80158015 This will always be `"assistant"`.

80168016

8017 - `"assistant"

8017

api/beta/sessions Changed · +15287 / -10074 lines

This page is larger than the 256 KiB this site keeps, so one side of the diff below stops where the stored text does.

Nothing in the body moved in this read. What changed is above.

api/beta/sessions/archive Changed · +255 / -43 lines

from line 307

307307

308308 - `configs: array of BetaManagedAgentsAgentToolConfig`

309309

310 - `enabled: boolean`

310 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

311311

312 - `name: "bash" or "edit" or "read" or 5 more`

312 Configuration for the bash tool.

313313

314 Built-in agent tool identifier.

314 - `enabled: boolean`

315315

316 - `"bash"`

316 - `name: "bash"`

317317

318 - `"edit"`

318 - `"bash"`

319319

320 - `"read"`

320 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

321321

322 - `"write"`

322 Permission policy for tool execution.

323323

324 - `"glob"`

324 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

325325

326 - `"grep"`

326 Tool calls are automatically approved without user confirmation.

327327

328 - `"web_fetch"`

328 - `type: "always_allow"`

329329

330 - `"web_search"`

330 - `"always_allow"`

331331

332 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

332 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

333333

334 Permission policy for tool execution.

334 Tool calls require user confirmation before execution.

335335

336 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

336 - `type: "always_ask"`

337337

338 Tool calls are automatically approved without user confirmation.

338 - `"always_ask"`

339339

340 - `type: "always_allow"`

340 - `type: "bash"`

341341

342 - `"always_allow"`

342 - `"bash"`

343343

344 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

344 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

345345

346 Tool calls require user confirmation before execution.

346 Configuration for the edit tool.

347347

348 - `type: "always_ask"`

348 - `enabled: boolean`

349349

350 - `"always_ask"`

350 - `name: "edit"`

351351

352 - `"edit"`

353

354 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

355

356 Permission policy for tool execution.

357

358 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

359

360 Tool calls are automatically approved without user confirmation.

361

362 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

363

364 Tool calls require user confirmation before execution.

365

366 - `type: "edit"`

367

368 - `"edit"`

369

370 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

371

372 Configuration for the read tool.

373

374 - `enabled: boolean`

375

376 - `name: "read"`

377

378 - `"read"`

379

380 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

381

382 Permission policy for tool execution.

383

384 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

385

386 Tool calls are automatically approved without user confirmation.

387

388 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

389

390 Tool calls require user confirmation before execution.

391

392 - `type: "read"`

393

394 - `"read"`

395

396 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

397

398 Configuration for the write tool.

399

400 - `enabled: boolean`

401

402 - `name: "write"`

403

404 - `"write"`

405

406 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

407

408 Permission policy for tool execution.

409

410 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

411

412 Tool calls are automatically approved without user confirmation.

413

414 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

415

416 Tool calls require user confirmation before execution.

417

418 - `type: "write"`

419

420 - `"write"`

421

422 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

423

424 Configuration for the glob tool.

425

426 - `enabled: boolean`

427

428 - `name: "glob"`

429

430 - `"glob"`

431

432 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

433

434 Permission policy for tool execution.

435

436 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

437

438 Tool calls are automatically approved without user confirmation.

439

440 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

441

442 Tool calls require user confirmation before execution.

443

444 - `type: "glob"`

445

446 - `"glob"`

447

448 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

449

450 Configuration for the grep tool.

451

452 - `enabled: boolean`

453

454 - `name: "grep"`

455

456 - `"grep"`

457

458 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

459

460 Permission policy for tool execution.

461

462 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

463

464 Tool calls are automatically approved without user confirmation.

465

466 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

467

468 Tool calls require user confirmation before execution.

469

470 - `type: "grep"`

471

472 - `"grep"`

473

474 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

475

476 Configuration for the web_fetch tool.

477

478 - `enabled: boolean`

479

480 - `name: "web_fetch"`

481

482 - `"web_fetch"`

483

484 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

485

486 Permission policy for tool execution.

487

488 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

489

490 Tool calls are automatically approved without user confirmation.

491

492 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

493

494 Tool calls require user confirmation before execution.

495

496 - `type: "web_fetch"`

497

498 - `"web_fetch"`

499

500 - `allowed_domains: optional array of string`

501

502 - `blocked_domains: optional array of string`

503

504 - `max_content_tokens: optional number or null`

505

506 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

507

508 Configuration for the web_search tool.

509

510 - `enabled: boolean`

511

512 - `name: "web_search"`

513

514 - `"web_search"`

515

516 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

517

518 Permission policy for tool execution.

519

520 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

521

522 Tool calls are automatically approved without user confirmation.

523

524 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

525

526 Tool calls require user confirmation before execution.

527

528 - `type: "web_search"`

529

530 - `"web_search"`

531

532 - `allowed_domains: optional array of string`

533

534 - `blocked_domains: optional array of string`

535

536 - `user_location: optional BetaManagedAgentsUserLocation or null`

537

538 Approximate user location for search result localization.

539

540 - `type: "approximate"`

541

542 Location precision. Only "approximate" is supported.

543

544 - `"approximate"`

545

546 - `city: optional string or null`

547

548 City name.

549

550 - `country: optional string or null`

551

552 Two-letter ISO 3166-1 country code, uppercase.

553

554 - `region: optional string or null`

555

556 Region or state name.

557

558 - `timezone: optional string or null`

559

560 IANA timezone identifier, e.g. "America/Los_Angeles".

561

352562 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

353563

354564 Resolved default configuration for agent tools.

from line 1027

8171027 "name": "bash",

8181028 "permission_policy": {

8191029 "type": "always_allow"

820 }

1030 },

1031 "type": "bash"

8211032 }

8221033 ],

8231034 "default_config": {

from line 1068

8571068 "name": "bash",

8581069 "permission_policy": {

8591070 "type": "always_allow"

860 }

1071 },

1072 "type": "bash"

8611073 }

8621074 ],

8631075 "default_config": {

api/beta/sessions/create Changed · +1140 / -678 lines

The two sides of this change are more than 400 edits apart, too far apart to line up, so this is the differ's own diff of it and the words inside a line are not marked.

from line 345

345345

346346 Per-tool configuration overrides.

347347

348 - `name: "bash" or "edit" or "read" or 5 more`

349

350 Built-in agent tool identifier.

351

352 - `"bash"`

353

354 - `"edit"`

355

356 - `"read"`

357

358 - `"write"`

359

360 - `"glob"`

361

362 - `"grep"`

363

364 - `"web_fetch"`

365

366 - `"web_search"`

348 - `BetaManagedAgentsBashToolConfigParams object { name, enabled, permission_policy, type }`

349

350 Configuration override for the bash tool.

351

352 - `name: "bash"`

353

354 Must be "bash".

355

356 - `"bash"`

357

358 - `enabled: optional boolean or null`

359

360 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

361

362 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

363

364 Permission policy for tool execution.

365

366 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

367

368 Tool calls are automatically approved without user confirmation.

369

370 - `type: "always_allow"`

371

372 - `"always_allow"`

373

374 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

375

376 Tool calls require user confirmation before execution.

377

378 - `type: "always_ask"`

379

380 - `"always_ask"`

381

382 - `type: optional "bash"`

383

384 - `"bash"`

385

386 - `BetaManagedAgentsEditToolConfigParams object { name, enabled, permission_policy, type }`

387

388 Configuration override for the edit tool.

389

390 - `name: "edit"`

391

392 Must be "edit".

393

394 - `"edit"`

395

396 - `enabled: optional boolean or null`

397

398 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

399

400 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

401

402 Permission policy for tool execution.

403

404 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

405

406 Tool calls are automatically approved without user confirmation.

407

408 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

409

410 Tool calls require user confirmation before execution.

411

412 - `type: optional "edit"`

413

414 - `"edit"`

415

416 - `BetaManagedAgentsReadToolConfigParams object { name, enabled, permission_policy, type }`

417

418 Configuration override for the read tool.

419

420 - `name: "read"`

421

422 Must be "read".

423

424 - `"read"`

425

426 - `enabled: optional boolean or null`

427

428 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

429

430 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

431

432 Permission policy for tool execution.

433

434 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

435

436 Tool calls are automatically approved without user confirmation.

437

438 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

439

440 Tool calls require user confirmation before execution.

441

442 - `type: optional "read"`

443

444 - `"read"`

445

446 - `BetaManagedAgentsWriteToolConfigParams object { name, enabled, permission_policy, type }`

447

448 Configuration override for the write tool.

449

450 - `name: "write"`

451

452 Must be "write".

453

454 - `"write"`

455

456 - `enabled: optional boolean or null`

457

458 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

459

460 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

461

462 Permission policy for tool execution.

463

464 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

465

466 Tool calls are automatically approved without user confirmation.

467

468 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

469

470 Tool calls require user confirmation before execution.

471

472 - `type: optional "write"`

473

474 - `"write"`

475

476 - `BetaManagedAgentsGlobToolConfigParams object { name, enabled, permission_policy, type }`

477

478 Configuration override for the glob tool.

479

480 - `name: "glob"`

481

482 Must be "glob".

483

484 - `"glob"`

485

486 - `enabled: optional boolean or null`

487

488 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

489

490 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

491

492 Permission policy for tool execution.

493

494 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

495

496 Tool calls are automatically approved without user confirmation.

497

498 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

499

500 Tool calls require user confirmation before execution.

501

502 - `type: optional "glob"`

503

504 - `"glob"`

505

506 - `BetaManagedAgentsGrepToolConfigParams object { name, enabled, permission_policy, type }`

507

508 Configuration override for the grep tool.

509

510 - `name: "grep"`

511

512 Must be "grep".

513

514 - `"grep"`

515

516 - `enabled: optional boolean or null`

517

518 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

519

520 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

521

522 Permission policy for tool execution.

523

524 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

525

526 Tool calls are automatically approved without user confirmation.

527

528 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

529

530 Tool calls require user confirmation before execution.

531

532 - `type: optional "grep"`

533

534 - `"grep"`

535

536 - `BetaManagedAgentsWebFetchToolConfigParams object { name, allowed_domains, blocked_domains, 4 more }`

537

538 Configuration override for the web_fetch tool.

539

540 - `name: "web_fetch"`

541

542 Must be "web_fetch".

543

544 - `"web_fetch"`

545

546 - `allowed_domains: optional array of string`

547

548 Only fetch URLs whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "docs.example.com" (no scheme, port, or path). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with blocked_domains.

549

550 - `blocked_domains: optional array of string`

551

552 Never fetch URLs whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "ads.example.com" (no scheme, port, or path). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with allowed_domains.

553

554 - `enabled: optional boolean or null`

555

556 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

557

558 - `max_content_tokens: optional number or null`

559

560 Maximum number of tokens of fetched text content to include in context per call. Does not apply to binary content such as PDFs.

561

562 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

563

564 Permission policy for tool execution.

565

566 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

567

568 Tool calls are automatically approved without user confirmation.

569

570 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

571

572 Tool calls require user confirmation before execution.

573

574 - `type: optional "web_fetch"`

575

576 - `"web_fetch"`

577

578 - `BetaManagedAgentsWebSearchToolConfigParams object { name, allowed_domains, blocked_domains, 4 more }`

579

580 Configuration override for the web_search tool.

581

582 - `name: "web_search"`

583

584 Must be "web_search".

585

586 - `"web_search"`

587

588 - `allowed_domains: optional array of string`

589

590 Only return search results whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "docs.example.com" (no scheme or port; an optional path suffix is accepted). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with blocked_domains.

591

592 - `blocked_domains: optional array of string`

593

594 Never return search results whose host is one of these domains or a subdomain of one. Each entry is a plain hostname like "ads.example.com" (no scheme or port; an optional path suffix is accepted). At most 64 entries; an empty list is rejected (omit the field instead). Cannot be combined with allowed_domains.

595

596 - `enabled: optional boolean or null`

597

598 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

599

600 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

601

602 Permission policy for tool execution.

603

604 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

605

606 Tool calls are automatically approved without user confirmation.

607

608 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

609

610 Tool calls require user confirmation before execution.

611

612 - `type: optional "web_search"`

613

614 - `"web_search"`

615

616 - `user_location: optional BetaManagedAgentsUserLocation or null`

617

618 Approximate user location for search result localization.

619

620 - `type: "approximate"`

621

622 Location precision. Only "approximate" is supported.

623

624 - `"approximate"`

625

626 - `city: optional string or null`

627

628 City name.

629

630 - `country: optional string or null`

631

632 Two-letter ISO 3166-1 country code, uppercase.

633

634 - `region: optional string or null`

635

636 Region or state name.

637

638 - `timezone: optional string or null`

639

640 IANA timezone identifier, e.g. "America/Los_Angeles".

641

642 - `default_config: optional BetaManagedAgentsAgentToolsetDefaultConfigParams or null`

643

644 Default configuration for all tools in a toolset.

367645

368646 - `enabled: optional boolean or null`

369647

370 Whether this tool is enabled and available to Claude. Overrides the default_config setting.

648 Whether tools are enabled and available to Claude by default. Defaults to true if not specified.

371649

372650 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

373651

from line 655

377655

378656 Tool calls are automatically approved without user confirmation.

379657

380 - `type: "always_allow"`

381

382 - `"always_allow"`

383

384658 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

385659

386660 Tool calls require user confirmation before execution.

387661

388 - `type: "always_ask"`

389

390 - `"always_ask"`

391

392 - `default_config: optional BetaManagedAgentsAgentToolsetDefaultConfigParams or null`

393

394 Default configuration for all tools in a toolset.

662 - `BetaManagedAgentsMCPToolsetParams object { mcp_server_name, type, configs, default_config }`

663

664 Configuration for tools from an MCP server defined in `mcp_servers`.

665

666 - `mcp_server_name: string`

667

668 Name of the MCP server. Must match a server name from the mcp_servers array. 1-255 characters.

669

670 - `type: "mcp_toolset"`

671

672 - `"mcp_toolset"`

673

674 - `configs: optional array of BetaManagedAgentsMCPToolConfigParams`

675

676 Per-tool configuration overrides.

677

678 - `name: string`

679

680 Name of the MCP tool to configure. 1-128 characters.

395681

396682 - `enabled: optional boolean or null`

397683

398 Whether tools are enabled and available to Claude by default. Defaults to true if not specified.

684 Whether this tool is enabled. Overrides the `default_config` setting.

399685

400686 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

401687

from line 695

409695

410696 Tool calls require user confirmation before execution.

411697

412 - `BetaManagedAgentsMCPToolsetParams object { mcp_server_name, type, configs, default_config }`

413

414 Configuration for tools from an MCP server defined in `mcp_servers`.

415

416 - `mcp_server_name: string`

417

418 Name of the MCP server. Must match a server name from the mcp_servers array. 1-255 characters.

419

420 - `type: "mcp_toolset"`

421

422 - `"mcp_toolset"`

423

424 - `configs: optional array of BetaManagedAgentsMCPToolConfigParams`

425

426 Per-tool configuration overrides.

698 - `default_config: optional BetaManagedAgentsMCPToolsetDefaultConfigParams or null`

699

700 Default configuration for all tools from an MCP server.

701

702 - `enabled: optional boolean or null`

703

704 Whether tools are enabled by default. Defaults to true if not specified.

705

706 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

707

708 Permission policy for tool execution.

709

710 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

711

712 Tool calls are automatically approved without user confirmation.

713

714 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

715

716 Tool calls require user confirmation before execution.

717

718 - `BetaManagedAgentsCustomToolParams object { description, input_schema, name, type }`

719

720 A custom tool that is executed by the API client rather than the agent. When the agent calls this tool, an `agent.custom_tool_use` event is emitted and the session goes idle, waiting for the client to provide the result via a `user.custom_tool_result` event.

721

722 - `description: string`

723

724 Description of what the tool does, shown to the agent to help it decide when to use the tool.

725

726 - `input_schema: BetaManagedAgentsCustomToolInputSchema`

727

728 JSON Schema for custom tool input parameters.

729

730 - `type: "object"`

731

732 - `"object"`

733

734 - `properties: optional map[unknown] or null`

735

736 - `required: optional array of string or null`

737

738 - `name: string`

739

740 Unique name for the tool. 1-128 characters; letters, digits, underscores, and hyphens.

741

742 - `type: "custom"`

743

744 - `"custom"`

745

746 - `version: optional number`

747

748 The specific `agent` version to use. Omit to use the latest version.

749

750- `environment_id: string`

751

752 ID of the `environment` defining the container configuration for this session.

753

754- `budget: optional BetaManagedAgentsBudgetLimit`

755

756 A hard spend ceiling. The session stops issuing new model requests once the tracked list cost reaches `max_list_cost`.

757

758 - `max_list_cost: BetaMonetaryAmount`

759

760 A monetary amount in a specific currency.

761

762 - `amount: string`

763

764 Amount in minor units of the currency, as an integer decimal string with no leading zeros: "2500" is $25.00 and "50" is fifty cents. A string rather than a number so no float rounding is ever applied.

765

766 - `currency: BetaCurrency`

767

768 Uppercase ISO-4217 currency code. `USD` is the only currency currently supported; the accepted set is closed and grows only when a new currency is priced.

769

770 - `"USD"`

771

772 - `type: "limit"`

773

774 - `"limit"`

775

776- `initial_events: optional array of BetaManagedAgentsUserMessageEventParams or BetaManagedAgentsUserDefineOutcomeEventParams`

777

778 Initial events to send to the `session` at creation, processed in order. Supports `user.message` and `user.define_outcome` events. Maximum 50 events.

779

780 - `BetaManagedAgentsUserMessageEventParams object { content, type }`

781

782 Parameters for sending a user message to the session.

783

784 - `content: array of BetaManagedAgentsTextBlock or BetaManagedAgentsImageBlock or BetaManagedAgentsDocumentBlock or BetaManagedAgentsRedactedBlock`

785

786 Array of content blocks for the user message.

787

788 - `BetaManagedAgentsTextBlock object { text, type }`

789

790 Regular text content.

791

792 - `text: string`

793

794 The text content.

795

796 - `type: "text"`

797

798 - `"text"`

799

800 - `BetaManagedAgentsImageBlock object { source, type }`

801

802 Image content specified directly as base64 data or as a reference via a URL.

803

804 - `source: BetaManagedAgentsBase64ImageSource or BetaManagedAgentsURLImageSource or BetaManagedAgentsFileImageSource`

805

806 Union type for image source variants.

807

808 - `BetaManagedAgentsBase64ImageSource object { data, media_type, type }`

809

810 Base64-encoded image data.

811

812 - `data: string`

813

814 Base64-encoded image data.

815

816 - `media_type: string`

817

818 MIME type of the image (e.g., "image/png", "image/jpeg", "image/gif", "image/webp").

819

820 - `type: "base64"`

821

822 - `"base64"`

823

824 - `BetaManagedAgentsURLImageSource object { type, url }`

825

826 Image referenced by URL.

827

828 - `type: "url"`

829

830 - `"url"`

831

832 - `url: string`

833

834 URL of the image to fetch.

835

836 - `BetaManagedAgentsFileImageSource object { file_id, type }`

837

838 Image referenced by file ID.

839

840 - `file_id: string`

841

842 ID of a previously uploaded file.

843

844 - `type: "file"`

845

846 - `"file"`

847

848 - `type: "image"`

849

850 - `"image"`

851

852 - `BetaManagedAgentsDocumentBlock object { source, type, context, title }`

853

854 Document content, either specified directly as base64 data, as text, or as a reference via a URL.

855

856 - `source: BetaManagedAgentsBase64DocumentSource or BetaManagedAgentsPlainTextDocumentSource or BetaManagedAgentsURLDocumentSource or BetaManagedAgentsFileDocumentSource`

857

858 Union type for document source variants.

859

860 - `BetaManagedAgentsBase64DocumentSource object { data, media_type, type }`

861

862 Base64-encoded document data.

863

864 - `data: string`

865

866 Base64-encoded document data.

867

868 - `media_type: string`

869

870 MIME type of the document (e.g., "application/pdf").

871

872 - `type: "base64"`

873

874 - `"base64"`

875

876 - `BetaManagedAgentsPlainTextDocumentSource object { data, media_type, type }`

877

878 Plain text document content.

879

880 - `data: string`

881

882 The plain text content.

883

884 - `media_type: "text/plain"`

885

886 MIME type of the text content. Must be "text/plain".

887

888 - `"text/plain"`

889

890 - `type: "text"`

891

892 - `"text"`

893

894 - `BetaManagedAgentsURLDocumentSource object { type, url }`

895

896 Document referenced by URL.

897

898 - `type: "url"`

899

900 - `"url"`

901

902 - `url: string`

903

904 URL of the document to fetch.

905

906 - `BetaManagedAgentsFileDocumentSource object { file_id, type }`

907

908 Document referenced by file ID.

909

910 - `file_id: string`

911

912 ID of a previously uploaded file.

913

914 - `type: "file"`

915

916 - `"file"`

917

918 - `type: "document"`

919

920 - `"document"`

921

922 - `context: optional string or null`

923

924 Additional context about the document for the model.

925

926 - `title: optional string or null`

927

928 The title of the document.

929

930 - `BetaManagedAgentsRedactedBlock object { type }`

931

932 Placeholder for content withheld by Anthropic model policy.

933

934 - `type: "redacted"`

935

936 - `"redacted"`

937

938 - `type: "user.message"`

939

940 - `"user.message"`

941

942 - `BetaManagedAgentsUserDefineOutcomeEventParams object { description, rubric, type, max_iterations }`

943

944 Parameters for defining an outcome the agent should work toward. The agent begins work on receipt.

945

946 - `description: string`

947

948 What the agent should produce. This is the task specification.

949

950 - `rubric: BetaManagedAgentsFileRubricParams or BetaManagedAgentsTextRubricParams`

951

952 Rubric for grading the quality of an outcome.

953

954 - `BetaManagedAgentsFileRubricParams object { file_id, type }`

955

956 Rubric referenced by a file uploaded via the Files API.

957

958 - `file_id: string`

959

960 ID of the rubric file.

961

962 - `type: "file"`

963

964 - `"file"`

965

966 - `BetaManagedAgentsTextRubricParams object { content, type }`

967

968 Rubric content provided inline as text.

969

970 - `content: string`

971

972 Rubric content. Plain text or markdown — the grader treats it as freeform text. Maximum 262144 characters.

973

974 - `type: "text"`

975

976 - `"text"`

977

978 - `type: "user.define_outcome"`

979

980 - `"user.define_outcome"`

981

982 - `max_iterations: optional number or null`

983

984 Eval→revision cycles before giving up. Default 3, max 20.

985

986- `metadata: optional map[string]`

987

988 Arbitrary key-value metadata attached to the session. Maximum 16 pairs, keys up to 64 chars, values up to 512 chars.

989

990- `resources: optional array of BetaManagedAgentsGitHubRepositoryResourceParams or BetaManagedAgentsFileResourceParams or BetaManagedAgentsMemoryStoreResourceParam`

991

992 Resources (e.g. repositories, files) to mount into the session's container.

993

994 - `BetaManagedAgentsGitHubRepositoryResourceParams object { authorization_token, type, url, 2 more }`

995

996 Mount a GitHub repository into the session's container.

997

998 - `authorization_token: string`

999

1000 GitHub authorization token used to clone the repository.

1001

1002 - `type: "github_repository"`

1003

1004 - `"github_repository"`

1005

1006 - `url: string`

1007

1008 Github URL of the repository

1009

1010 - `checkout: optional BetaManagedAgentsBranchCheckout or BetaManagedAgentsCommitCheckout or null`

1011

1012 Branch or commit to check out. Defaults to the repository's default branch.

1013

1014 - `BetaManagedAgentsBranchCheckout object { name, type }`

1015

1016 - `name: string`

1017

1018 Branch name to check out.

1019

1020 - `type: "branch"`

1021

1022 - `"branch"`

1023

1024 - `BetaManagedAgentsCommitCheckout object { sha, type }`

1025

1026 - `sha: string`

1027

1028 Full commit SHA to check out.

1029

1030 - `type: "commit"`

1031

1032 - `"commit"`

1033

1034 - `mount_path: optional string or null`

1035

1036 Mount path in the container. Defaults to `/workspace/<repo-name>`.

1037

1038 - `BetaManagedAgentsFileResourceParams object { file_id, type, mount_path }`

1039

1040 Mount a file uploaded via the Files API into the session.

1041

1042 - `file_id: string`

1043

1044 ID of a previously uploaded file.

1045

1046 - `type: "file"`

1047

1048 - `"file"`

1049

1050 - `mount_path: optional string or null`

1051

1052 Mount path in the container. Defaults to `/mnt/session/uploads/<file_id>`.

1053

1054 - `BetaManagedAgentsMemoryStoreResourceParam object { memory_store_id, type, access, instructions }`

1055

1056 Parameters for attaching a memory store to an agent session.

1057

1058 - `memory_store_id: string`

1059

1060 The memory store ID (memstore_...). Must belong to the caller's organization and workspace.

1061

1062 - `type: "memory_store"`

1063

1064 - `"memory_store"`

1065

1066 - `access: optional "read_write" or "read_only" or null`

1067

1068 Access mode for an attached memory store.

1069

1070 - `"read_write"`

1071

1072 - `"read_only"`

1073

1074 - `instructions: optional string or null`

1075

1076 Per-attachment guidance for the agent on how to use this store. Rendered into the memory section of the system prompt. Max 4096 chars.

1077

1078- `title: optional string or null`

1079

1080 Human-readable session title.

1081

1082- `vault_ids: optional array of string`

1083

1084 Vault IDs for stored credentials the agent can use during the session.

1085

1086### Returns

1087

1088- `BetaManagedAgentsSession object { id, agent, archived_at, 14 more }`

1089

1090 A Managed Agents `session`.

1091

1092 - `id: string`

1093

1094 - `agent: BetaManagedAgentsSessionAgent`

1095

1096 Resolved `agent` definition for a `session`. Snapshot of the `agent` at `session` creation time.

1097

1098 - `id: string`

1099

1100 - `description: string or null`

1101

1102 - `mcp_servers: array of BetaManagedAgentsMCPServerURLDefinition`

1103

1104 - `name: string`

1105

1106 - `type: "url"`

1107

1108 - `"url"`

1109

1110 - `url: string`

1111

1112 - `model: BetaManagedAgentsModelConfig`

1113

1114 Model identifier and configuration.

1115

1116 - `id: BetaManagedAgentsModel`

1117

1118 The model that will power your agent.

1119

1120 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.

1121

1122 - `"claude-sonnet-5" or "claude-fable-5" or "claude-opus-5" or 10 more`

1123

1124 The model that will power your agent.

1125

1126 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.

1127

1128 - `"claude-sonnet-5"`

1129

1130 High-performance model for coding and agents

1131

1132 - `"claude-fable-5"`

1133

1134 Next generation of intelligence for the hardest knowledge work and coding problems

1135

1136 - `"claude-opus-5"`

1137

1138 Powerful intelligence for long-running agents and coding

1139

1140 - `"claude-opus-4-8"`

1141

1142 Powerful intelligence for long-running agents and coding

1143

1144 - `"claude-opus-4-7"`

1145

1146 Powerful intelligence for long-running agents and coding

1147

1148 - `"claude-opus-4-6"`

1149

1150 Powerful intelligence for long-running agents and coding

1151

1152 - `"claude-sonnet-4-6"`

1153

1154 Best combination of speed and intelligence

1155

1156 - `"claude-haiku-4-5"`

1157

1158 Fastest model with near-frontier intelligence

1159

1160 - `"claude-haiku-4-5-20251001"`

1161

1162 Fastest model with near-frontier intelligence

1163

1164 - `"claude-opus-4-5"`

1165

1166 Powerful intelligence for long-running agents and coding

1167

1168 - `"claude-opus-4-5-20251101"`

1169

1170 Powerful intelligence for long-running agents and coding

1171

1172 - `"claude-sonnet-4-5"`

1173

1174 High-performance model for agents and coding

1175

1176 - `"claude-sonnet-4-5-20250929"`

1177

1178 High-performance model for agents and coding

1179

1180 - `string`

1181

1182 - `effort: optional BetaManagedAgentsEffortLow or BetaManagedAgentsEffortMedium or BetaManagedAgentsEffortHigh or 2 more`

1183

1184 How hard Claude works on each turn. Sets `output_config.effort` on every Messages call the session makes.

1185

1186 - `BetaManagedAgentsEffortLow object { type }`

1187

1188 Low effort. Favors latency over reasoning depth.

1189

1190 - `type: "low"`

1191

1192 - `"low"`

1193

1194 - `BetaManagedAgentsEffortMedium object { type }`

1195

1196 Medium effort. Balances latency and reasoning depth.

1197

1198 - `type: "medium"`

1199

1200 - `"medium"`

1201

1202 - `BetaManagedAgentsEffortHigh object { type }`

1203

1204 High effort. Favors reasoning depth.

1205

1206 - `type: "high"`

1207

1208 - `"high"`

1209

1210 - `BetaManagedAgentsEffortXhigh object { type }`

1211

1212 Extra-high effort. Not all models accept this level.

1213

1214 - `type: "xhigh"`

1215

1216 - `"xhigh"`

1217

1218 - `BetaManagedAgentsEffortMax object { type }`

1219

1220 Maximum effort. Favors reasoning depth over latency.

1221

1222 - `type: "max"`

1223

1224 - `"max"`

1225

1226 - `inference_geo: optional string`

1227

1228 Geographic region for model inference. When unset, requests fall through to the workspace's default_inference_geo.

1229

1230 - `speed: optional "standard" or "fast"`

1231

1232 Inference speed mode. `fast` provides significantly faster output token generation at premium pricing. Not all models support `fast`; invalid combinations are rejected at create time.

1233

1234 - `"standard"`

1235

1236 - `"fast"`

1237

1238 - `multiagent: BetaManagedAgentsSessionMultiagentCoordinator or null`

1239

1240 Resolved coordinator topology with full agent definitions for each roster member.

1241

1242 - `agents: array of BetaManagedAgentsSessionThreadAgent or BetaManagedAgentsAdvisor`

1243

1244 Full `agent` definitions the coordinator may spawn as session threads.

1245

1246 - `BetaManagedAgentsSessionThreadAgent object { id, description, mcp_servers, 7 more }`

1247

1248 Resolved `agent` definition for a single `session_thread`. Snapshot of the agent at thread creation time. The multiagent roster is not repeated here; read it from `Session.agent`.

1249

1250 - `id: string`

1251

1252 - `description: string or null`

1253

1254 - `mcp_servers: array of BetaManagedAgentsMCPServerURLDefinition`

1255

1256 - `name: string`

1257

1258 - `type: "url"`

1259

1260 - `url: string`

1261

1262 - `model: BetaManagedAgentsModelConfig`

1263

1264 Model identifier and configuration.

4271265

4281266 - `name: string`

4291267

430 Name of the MCP tool to configure. 1-128 characters.

431

432 - `enabled: optional boolean or null`

433

434 Whether this tool is enabled. Overrides the `default_config` setting.

435

436 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

437

438 Permission policy for tool execution.

439

440 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

441

442 Tool calls are automatically approved without user confirmation.

443

444 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

445

446 Tool calls require user confirmation before execution.

447

448 - `default_config: optional BetaManagedAgentsMCPToolsetDefaultConfigParams or null`

449

450 Default configuration for all tools from an MCP server.

451

452 - `enabled: optional boolean or null`

453

454 Whether tools are enabled by default. Defaults to true if not specified.

455

456 - `permission_policy: optional BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy or null`

457

458 Permission policy for tool execution.

459

460 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

461

462 Tool calls are automatically approved without user confirmation.

463

464 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

465

466 Tool calls require user confirmation before execution.

467

468 - `BetaManagedAgentsCustomToolParams object { description, input_schema, name, type }`

469

470 A custom tool that is executed by the API client rather than the agent. When the agent calls this tool, an `agent.custom_tool_use` event is emitted and the session goes idle, waiting for the client to provide the result via a `user.custom_tool_result` event.

471

472 - `description: string`

473

474 Description of what the tool does, shown to the agent to help it decide when to use the tool.

475

476 - `input_schema: BetaManagedAgentsCustomToolInputSchema`

477

478 JSON Schema for custom tool input parameters.

479

480 - `type: "object"`

481

482 - `"object"`

483

484 - `properties: optional map[unknown] or null`

485

486 - `required: optional array of string or null`

487

488 - `name: string`

489

490 Unique name for the tool. 1-128 characters; letters, digits, underscores, and hyphens.

491

492 - `type: "custom"`

493

494 - `"custom"`

495

496 - `version: optional number`

497

498 The specific `agent` version to use. Omit to use the latest version.

499

500- `environment_id: string`

501

502 ID of the `environment` defining the container configuration for this session.

503

504- `budget: optional BetaManagedAgentsBudgetLimit`

505

506 A hard spend ceiling. The session stops issuing new model requests once the tracked list cost reaches `max_list_cost`.

507

508 - `max_list_cost: BetaMonetaryAmount`

509

510 A monetary amount in a specific currency.

511

512 - `amount: string`

513

514 Amount in minor units of the currency, as an integer decimal string with no leading zeros: "2500" is $25.00 and "50" is fifty cents. A string rather than a number so no float rounding is ever applied.

515

516 - `currency: BetaCurrency`

517

518 Uppercase ISO-4217 currency code. `USD` is the only currency currently supported; the accepted set is closed and grows only when a new currency is priced.

519

520 - `"USD"`

521

522 - `type: "limit"`

523

524 - `"limit"`

525

526- `initial_events: optional array of BetaManagedAgentsUserMessageEventParams or BetaManagedAgentsUserDefineOutcomeEventParams`

527

528 Initial events to send to the `session` at creation, processed in order. Supports `user.message` and `user.define_outcome` events. Maximum 50 events.

529

530 - `BetaManagedAgentsUserMessageEventParams object { content, type }`

531

532 Parameters for sending a user message to the session.

533

534 - `content: array of BetaManagedAgentsTextBlock or BetaManagedAgentsImageBlock or BetaManagedAgentsDocumentBlock or BetaManagedAgentsRedactedBlock`

535

536 Array of content blocks for the user message.

537

538 - `BetaManagedAgentsTextBlock object { text, type }`

539

540 Regular text content.

541

542 - `text: string`

543

544 The text content.

545

546 - `type: "text"`

547

548 - `"text"`

549

550 - `BetaManagedAgentsImageBlock object { source, type }`

551

552 Image content specified directly as base64 data or as a reference via a URL.

553

554 - `source: BetaManagedAgentsBase64ImageSource or BetaManagedAgentsURLImageSource or BetaManagedAgentsFileImageSource`

555

556 Union type for image source variants.

557

558 - `BetaManagedAgentsBase64ImageSource object { data, media_type, type }`

559

560 Base64-encoded image data.

561

562 - `data: string`

563

564 Base64-encoded image data.

565

566 - `media_type: string`

567

568 MIME type of the image (e.g., "image/png", "image/jpeg", "image/gif", "image/webp").

569

570 - `type: "base64"`

571

572 - `"base64"`

573

574 - `BetaManagedAgentsURLImageSource object { type, url }`

575

576 Image referenced by URL.

577

578 - `type: "url"`

579

580 - `"url"`

581

582 - `url: string`

583

584 URL of the image to fetch.

585

586 - `BetaManagedAgentsFileImageSource object { file_id, type }`

587

588 Image referenced by file ID.

589

590 - `file_id: string`

591

592 ID of a previously uploaded file.

593

594 - `type: "file"`

595

596 - `"file"`

597

598 - `type: "image"`

599

600 - `"image"`

601

602 - `BetaManagedAgentsDocumentBlock object { source, type, context, title }`

603

604 Document content, either specified directly as base64 data, as text, or as a reference via a URL.

605

606 - `source: BetaManagedAgentsBase64DocumentSource or BetaManagedAgentsPlainTextDocumentSource or BetaManagedAgentsURLDocumentSource or BetaManagedAgentsFileDocumentSource`

607

608 Union type for document source variants.

609

610 - `BetaManagedAgentsBase64DocumentSource object { data, media_type, type }`

611

612 Base64-encoded document data.

613

614 - `data: string`

615

616 Base64-encoded document data.

617

618 - `media_type: string`

619

620 MIME type of the document (e.g., "application/pdf").

621

622 - `type: "base64"`

623

624 - `"base64"`

625

626 - `BetaManagedAgentsPlainTextDocumentSource object { data, media_type, type }`

627

628 Plain text document content.

629

630 - `data: string`

631

632 The plain text content.

633

634 - `media_type: "text/plain"`

635

636 MIME type of the text content. Must be "text/plain".

637

638 - `"text/plain"`

639

640 - `type: "text"`

641

642 - `"text"`

643

644 - `BetaManagedAgentsURLDocumentSource object { type, url }`

645

646 Document referenced by URL.

647

648 - `type: "url"`

649

650 - `"url"`

651

652 - `url: string`

653

654 URL of the document to fetch.

655

656 - `BetaManagedAgentsFileDocumentSource object { file_id, type }`

657

658 Document referenced by file ID.

659

660 - `file_id: string`

661

662 ID of a previously uploaded file.

663

664 - `type: "file"`

665

666 - `"file"`

667

668 - `type: "document"`

669

670 - `"document"`

671

672 - `context: optional string or null`

673

674 Additional context about the document for the model.

675

676 - `title: optional string or null`

677

678 The title of the document.

679

680 - `BetaManagedAgentsRedactedBlock object { type }`

681

682 Placeholder for content withheld by Anthropic model policy.

683

684 - `type: "redacted"`

685

686 - `"redacted"`

687

688 - `type: "user.message"`

689

690 - `"user.message"`

691

692 - `BetaManagedAgentsUserDefineOutcomeEventParams object { description, rubric, type, max_iterations }`

693

694 Parameters for defining an outcome the agent should work toward. The agent begins work on receipt.

695

696 - `description: string`

697

698 What the agent should produce. This is the task specification.

699

700 - `rubric: BetaManagedAgentsFileRubricParams or BetaManagedAgentsTextRubricParams`

701

702 Rubric for grading the quality of an outcome.

703

704 - `BetaManagedAgentsFileRubricParams object { file_id, type }`

705

706 Rubric referenced by a file uploaded via the Files API.

707

708 - `file_id: string`

709

710 ID of the rubric file.

711

712 - `type: "file"`

713

714 - `"file"`

715

716 - `BetaManagedAgentsTextRubricParams object { content, type }`

717

718 Rubric content provided inline as text.

719

720 - `content: string`

721

722 Rubric content. Plain text or markdown — the grader treats it as freeform text. Maximum 262144 characters.

723

724 - `type: "text"`

725

726 - `"text"`

727

728 - `type: "user.define_outcome"`

729

730 - `"user.define_outcome"`

731

732 - `max_iterations: optional number or null`

733

734 Eval→revision cycles before giving up. Default 3, max 20.

735

736- `metadata: optional map[string]`

737

738 Arbitrary key-value metadata attached to the session. Maximum 16 pairs, keys up to 64 chars, values up to 512 chars.

739

740- `resources: optional array of BetaManagedAgentsGitHubRepositoryResourceParams or BetaManagedAgentsFileResourceParams or BetaManagedAgentsMemoryStoreResourceParam`

741

742 Resources (e.g. repositories, files) to mount into the session's container.

743

744 - `BetaManagedAgentsGitHubRepositoryResourceParams object { authorization_token, type, url, 2 more }`

745

746 Mount a GitHub repository into the session's container.

747

748 - `authorization_token: string`

749

750 GitHub authorization token used to clone the repository.

751

752 - `type: "github_repository"`

753

754 - `"github_repository"`

755

756 - `url: string`

757

758 Github URL of the repository

759

760 - `checkout: optional BetaManagedAgentsBranchCheckout or BetaManagedAgentsCommitCheckout or null`

761

762 Branch or commit to check out. Defaults to the repository's default branch.

763

764 - `BetaManagedAgentsBranchCheckout object { name, type }`

765

766 - `name: string`

767

768 Branch name to check out.

769

770 - `type: "branch"`

771

772 - `"branch"`

773

774 - `BetaManagedAgentsCommitCheckout object { sha, type }`

775

776 - `sha: string`

777

778 Full commit SHA to check out.

779

780 - `type: "commit"`

781

782 - `"commit"`

783

784 - `mount_path: optional string or null`

785

786 Mount path in the container. Defaults to `/workspace/<repo-name>`.

787

788 - `BetaManagedAgentsFileResourceParams object { file_id, type, mount_path }`

789

790 Mount a file uploaded via the Files API into the session.

791

792 - `file_id: string`

793

794 ID of a previously uploaded file.

795

796 - `type: "file"`

797

798 - `"file"`

799

800 - `mount_path: optional string or null`

801

802 Mount path in the container. Defaults to `/mnt/session/uploads/<file_id>`.

803

804 - `BetaManagedAgentsMemoryStoreResourceParam object { memory_store_id, type, access, instructions }`

805

806 Parameters for attaching a memory store to an agent session.

807

808 - `memory_store_id: string`

809

810 The memory store ID (memstore_...). Must belong to the caller's organization and workspace.

811

812 - `type: "memory_store"`

813

814 - `"memory_store"`

815

816 - `access: optional "read_write" or "read_only" or null`

817

818 Access mode for an attached memory store.

819

820 - `"read_write"`

821

822 - `"read_only"`

823

824 - `instructions: optional string or null`

825

826 Per-attachment guidance for the agent on how to use this store. Rendered into the memory section of the system prompt. Max 4096 chars.

827

828- `title: optional string or null`

829

830 Human-readable session title.

831

832- `vault_ids: optional array of string`

833

834 Vault IDs for stored credentials the agent can use during the session.

835

836### Returns

837

838- `BetaManagedAgentsSession object { id, agent, archived_at, 14 more }`

839

840 A Managed Agents `session`.

841

842 - `id: string`

843

844 - `agent: BetaManagedAgentsSessionAgent`

845

846 Resolved `agent` definition for a `session`. Snapshot of the `agent` at `session` creation time.

847

848 - `id: string`

849

850 - `description: string or null`

851

852 - `mcp_servers: array of BetaManagedAgentsMCPServerURLDefinition`

853

854 - `name: string`

855

856 - `type: "url"`

857

858 - `"url"`

859

860 - `url: string`

861

862 - `model: BetaManagedAgentsModelConfig`

863

864 Model identifier and configuration.

865

866 - `id: BetaManagedAgentsModel`

867

868 The model that will power your agent.

869

870 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.

871

872 - `"claude-sonnet-5" or "claude-fable-5" or "claude-opus-5" or 10 more`

873

874 The model that will power your agent.

875

876 See [models](https://docs.anthropic.com/en/docs/models-overview) for additional details and options.

877

878 - `"claude-sonnet-5"`

879

880 High-performance model for coding and agents

881

882 - `"claude-fable-5"`

883

884 Next generation of intelligence for the hardest knowledge work and coding problems

885

886 - `"claude-opus-5"`

887

888 Powerful intelligence for long-running agents and coding

889

890 - `"claude-opus-4-8"`

891

892 Powerful intelligence for long-running agents and coding

893

894 - `"claude-opus-4-7"`

895

896 Powerful intelligence for long-running agents and coding

897

898 - `"claude-opus-4-6"`

899

900 Powerful intelligence for long-running agents and coding

901

902 - `"claude-sonnet-4-6"`

903

904 Best combination of speed and intelligence

905

906 - `"claude-haiku-4-5"`

907

908 Fastest model with near-frontier intelligence

909

910 - `"claude-haiku-4-5-20251001"`

911

912 Fastest model with near-frontier intelligence

913

914 - `"claude-opus-4-5"`

915

916 Powerful intelligence for long-running agents and coding

917

918 - `"claude-opus-4-5-20251101"`

919

920 Powerful intelligence for long-running agents and coding

921

922 - `"claude-sonnet-4-5"`

923

924 High-performance model for agents and coding

925

926 - `"claude-sonnet-4-5-20250929"`

927

928 High-performance model for agents and coding

929

930 - `string`

931

932 - `effort: optional BetaManagedAgentsEffortLow or BetaManagedAgentsEffortMedium or BetaManagedAgentsEffortHigh or 2 more`

933

934 How hard Claude works on each turn. Sets `output_config.effort` on every Messages call the session makes.

935

936 - `BetaManagedAgentsEffortLow object { type }`

937

938 Low effort. Favors latency over reasoning depth.

939

940 - `type: "low"`

941

942 - `"low"`

943

944 - `BetaManagedAgentsEffortMedium object { type }`

945

946 Medium effort. Balances latency and reasoning depth.

947

948 - `type: "medium"`

949

950 - `"medium"`

951

952 - `BetaManagedAgentsEffortHigh object { type }`

953

954 High effort. Favors reasoning depth.

955

956 - `type: "high"`

957

958 - `"high"`

959

960 - `BetaManagedAgentsEffortXhigh object { type }`

961

962 Extra-high effort. Not all models accept this level.

963

964 - `type: "xhigh"`

965

966 - `"xhigh"`

967

968 - `BetaManagedAgentsEffortMax object { type }`

969

970 Maximum effort. Favors reasoning depth over latency.

971

972 - `type: "max"`

973

974 - `"max"`

975

976 - `inference_geo: optional string`

977

978 Geographic region for model inference. When unset, requests fall through to the workspace's default_inference_geo.

979

980 - `speed: optional "standard" or "fast"`

981

982 Inference speed mode. `fast` provides significantly faster output token generation at premium pricing. Not all models support `fast`; invalid combinations are rejected at create time.

983

984 - `"standard"`

985

986 - `"fast"`

987

988 - `multiagent: BetaManagedAgentsSessionMultiagentCoordinator or null`

989

990 Resolved coordinator topology with full agent definitions for each roster member.

991

992 - `agents: array of BetaManagedAgentsSessionThreadAgent or BetaManagedAgentsAdvisor`

993

994 Full `agent` definitions the coordinator may spawn as session threads.

995

996 - `BetaManagedAgentsSessionThreadAgent object { id, description, mcp_servers, 7 more }`

997

998 Resolved `agent` definition for a single `session_thread`. Snapshot of the agent at thread creation time. The multiagent roster is not repeated here; read it from `Session.agent`.

999

1000 - `id: string`

1001

1002 - `description: string or null`

1003

1004 - `mcp_servers: array of BetaManagedAgentsMCPServerURLDefinition`

1005

1006 - `name: string`

1007

1008 - `type: "url"`

1009

1010 - `url: string`

1011

1012 - `model: BetaManagedAgentsModelConfig`

1013

1014 Model identifier and configuration.

1015

1016 - `name: string`

1017

10181268 - `skills: array of BetaManagedAgentsAnthropicSkill or BetaManagedAgentsCustomSkill`

10191269

10201270 - `BetaManagedAgentsAnthropicSkill object { skill_id, type, version }`

from line 1299

10491299

10501300 - `configs: array of BetaManagedAgentsAgentToolConfig`

10511301

1052 - `enabled: boolean`

1053

1054 - `name: "bash" or "edit" or "read" or 5 more`

1055

1056 Built-in agent tool identifier.

1057

1058 - `"bash"`

1059

1060 - `"edit"`

1061

1062 - `"read"`

1063

1064 - `"write"`

1065

1066 - `"glob"`

1067

1068 - `"grep"`

1069

1070 - `"web_fetch"`

1071

1072 - `"web_search"`

1073

1074 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1075

1076 Permission policy for tool execution.

1077

1078 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1079

1080 Tool calls are automatically approved without user confirmation.

1081

1082 - `type: "always_allow"`

1083

1084 - `"always_allow"`

1085

1086 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1087

1088 Tool calls require user confirmation before execution.

1089

1090 - `type: "always_ask"`

1091

1092 - `"always_ask"`

1302 - `BetaManagedAgentsBashToolConfig object { enabled, name, permission_policy, type }`

1303

1304 Configuration for the bash tool.

1305

1306 - `enabled: boolean`

1307

1308 - `name: "bash"`

1309

1310 - `"bash"`

1311

1312 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1313

1314 Permission policy for tool execution.

1315

1316 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1317

1318 Tool calls are automatically approved without user confirmation.

1319

1320 - `type: "always_allow"`

1321

1322 - `"always_allow"`

1323

1324 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1325

1326 Tool calls require user confirmation before execution.

1327

1328 - `type: "always_ask"`

1329

1330 - `"always_ask"`

1331

1332 - `type: "bash"`

1333

1334 - `"bash"`

1335

1336 - `BetaManagedAgentsEditToolConfig object { enabled, name, permission_policy, type }`

1337

1338 Configuration for the edit tool.

1339

1340 - `enabled: boolean`

1341

1342 - `name: "edit"`

1343

1344 - `"edit"`

1345

1346 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1347

1348 Permission policy for tool execution.

1349

1350 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1351

1352 Tool calls are automatically approved without user confirmation.

1353

1354 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1355

1356 Tool calls require user confirmation before execution.

1357

1358 - `type: "edit"`

1359

1360 - `"edit"`

1361

1362 - `BetaManagedAgentsReadToolConfig object { enabled, name, permission_policy, type }`

1363

1364 Configuration for the read tool.

1365

1366 - `enabled: boolean`

1367

1368 - `name: "read"`

1369

1370 - `"read"`

1371

1372 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1373

1374 Permission policy for tool execution.

1375

1376 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1377

1378 Tool calls are automatically approved without user confirmation.

1379

1380 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1381

1382 Tool calls require user confirmation before execution.

1383

1384 - `type: "read"`

1385

1386 - `"read"`

1387

1388 - `BetaManagedAgentsWriteToolConfig object { enabled, name, permission_policy, type }`

1389

1390 Configuration for the write tool.

1391

1392 - `enabled: boolean`

1393

1394 - `name: "write"`

1395

1396 - `"write"`

1397

1398 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1399

1400 Permission policy for tool execution.

1401

1402 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1403

1404 Tool calls are automatically approved without user confirmation.

1405

1406 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1407

1408 Tool calls require user confirmation before execution.

1409

1410 - `type: "write"`

1411

1412 - `"write"`

1413

1414 - `BetaManagedAgentsGlobToolConfig object { enabled, name, permission_policy, type }`

1415

1416 Configuration for the glob tool.

1417

1418 - `enabled: boolean`

1419

1420 - `name: "glob"`

1421

1422 - `"glob"`

1423

1424 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1425

1426 Permission policy for tool execution.

1427

1428 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1429

1430 Tool calls are automatically approved without user confirmation.

1431

1432 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1433

1434 Tool calls require user confirmation before execution.

1435

1436 - `type: "glob"`

1437

1438 - `"glob"`

1439

1440 - `BetaManagedAgentsGrepToolConfig object { enabled, name, permission_policy, type }`

1441

1442 Configuration for the grep tool.

1443

1444 - `enabled: boolean`

1445

1446 - `name: "grep"`

1447

1448 - `"grep"`

1449

1450 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1451

1452 Permission policy for tool execution.

1453

1454 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1455

1456 Tool calls are automatically approved without user confirmation.

1457

1458 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1459

1460 Tool calls require user confirmation before execution.

1461

1462 - `type: "grep"`

1463

1464 - `"grep"`

1465

1466 - `BetaManagedAgentsWebFetchToolConfig object { enabled, name, permission_policy, 4 more }`

1467

1468 Configuration for the web_fetch tool.

1469

1470 - `enabled: boolean`

1471

1472 - `name: "web_fetch"`

1473

1474 - `"web_fetch"`

1475

1476 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1477

1478 Permission policy for tool execution.

1479

1480 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1481

1482 Tool calls are automatically approved without user confirmation.

1483

1484 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1485

1486 Tool calls require user confirmation before execution.

1487

1488 - `type: "web_fetch"`

1489

1490 - `"web_fetch"`

1491

1492 - `allowed_domains: optional array of string`

1493

1494 - `blocked_domains: optional array of string`

1495

1496 - `max_content_tokens: optional number or null`

1497

1498 - `BetaManagedAgentsWebSearchToolConfig object { enabled, name, permission_policy, 4 more }`

1499

1500 Configuration for the web_search tool.

1501

1502 - `enabled: boolean`

1503

1504 - `name: "web_search"`

1505

1506 - `"web_search"`

1507

1508 - `permission_policy: BetaManagedAgentsAlwaysAllowPolicy or BetaManagedAgentsAlwaysAskPolicy`

1509

1510 Permission policy for tool execution.

1511

1512 - `BetaManagedAgentsAlwaysAllowPolicy object { type }`

1513

1514 Tool calls are automatically approved without user confirmation.

1515

1516 - `BetaManagedAgentsAlwaysAskPolicy object { type }`

1517

1518 Tool calls require user confirmation before execution.

1519

1520 - `type: "web_search"`

1521

1522 - `"web_search"`

1523

1524 - `allowed_domains: optional array of string`

1525

1526 - `blocked_domains: optional array of string`

1527

1528 - `user_location: optional BetaManagedAgentsUserLocation or null`

1529

1530 Approximate user location for search result localization.

1531

1532 - `type: "approximate"`

1533

1534 Location precision. Only "approximate" is supported.

1535

1536 - `"approximate"`

1537

1538 - `city: optional string or null`

1539

1540 City name.

1541

1542 - `country: optional string or null`

1543

1544 Two-letter ISO 3166-1 country code, uppercase.

1545

1546 - `region: optional string or null`

1547

1548 Region or state name.

1549

1550 - `timezone: optional string or null`

1551

1552 IANA timezone identifier, e.g. "America/Los_Angeles".

10931553

10941554 - `default_config: BetaManagedAgentsAgentToolsetDefaultConfig`

10951555

from line 2024

15642024 "name": "bash",

15652025 "permission_policy": {

15662026 "type": "always_allow"

1567 }

2027 },

2028 "type": "bash"

15682029 }

15692030 ],

15702031 "default_config": {

from line 2065

16042065 "name": "bash",

16052066 "permission_policy": {

16062067 "type": "always_allow"

1607 }

2068 },

2069 "type": "bash"

16082070 }

16092071 ],

16102072 "default_config": {

16112073