An entry is only as true as the day it was written.
Flags move on the server after a release ships. Anthropic's own notes land days later and their documentation later still. Sometimes an entry turns out to be wrong and gets corrected by hand. All of that is recorded against the entry it happened to, and this is the whole of it in the order it happened.
Nothing here replaces an entry. Every line links to the page its release published, still saying what that release said, because a correction is added and dated rather than swapped in.
| When | Entry | What happened |
|---|---|---|
v2.1.274agent.register |
Documentation arrivedAnthropic's documentation has since written up agent.register, on Claude Code changelog.
the line in the docs* Fixed `$.agent.register` succeeding from a mod's hook that was still running after the mod was reloaded or removed; the call is now refused |
|
v2.1.295forceLoginMethod |
Documentation arrivedAnthropic's documentation has since written up forceLoginMethod, on Claude Code changelog.
the line in the docs* Fixed a saved Claude apps gateway sign-in being ignored on machines whose managed settings set `forceLoginMethod` to `gateway` with no `forceLoginGatewayUrl` (regression in 2.1.295) |
|
v2.1.287CLAUDE_CODE_TRANSCRIPT_LOCAL_GC |
Documentation arrivedAnthropic's documentation has since written up CLAUDE_CODE_TRANSCRIPT_LOCAL_GC, on Claude Code changelog.
the line in the docs- Improved `CLAUDE_CODE_TRANSCRIPT_LOCAL_GC`: a large transcript file is no longer rewritten when that would free under 10% of it |
|
v2.1.288autoCompactWindow |
Documentation arrivedAnthropic's documentation has since written up autoCompactWindow, on Claude Code changelog.
the line in the docs* Added `autoCompactWindow` to subagent frontmatter and `--agents` definitions, so a subagent can auto-compact earlier than the main conversation's window |
|
v2.1.288autoCompactWindow |
Documentation arrivedAnthropic's documentation has since written up autoCompactWindow, on Claude Code changelog.
the line in the docs* Added `autoCompactWindow` to subagent frontmatter and `--agents` definitions, so a subagent can auto-compact earlier than the main conversation's window |
|
v2.1.295claude plugin validate |
Documentation arrivedAnthropic's documentation has since written up claude plugin validate, on Troubleshoot a mod.
the line in the docsThe line starts with the mod's name, then `hooks module did not load:`, the file, and `code nested too deep to scan: more than 2000 scopes`. A file in a hooks module can't nest scopes, such as functions, blocks, and loops, more than [2,000… |
|
v2.1.293isDeferred |
Documentation arrivedAnthropic's documentation has since written up isDeferred, on Mods reference.
the line in the docs| `tool.describe` | Once for each tool, when its description is first sent to Claude. A second time for an MCP tool when Claude loads it through [tool search](/docs/en/mcp#scale-with-mcp-tool-search), with `e.description` set to the text C… |
|
v2.1.293isDeferred |
Documentation arrivedAnthropic's documentation has since written up isDeferred, on Mods reference.
the line in the docs| `tool.describe` | Once for each tool, when its description is first sent to Claude. A second time for an MCP tool when Claude loads it through [tool search](/docs/en/mcp#scale-with-mcp-tool-search), with `e.description` set to the text C… |
|
v2.1.295claude plugin validate |
Documentation arrivedA small documentation edit on Marketplace reference touched a line naming claude plugin validate after this was published.
the line in the docs| `name` | string | Marketplace identifier: letters, digits, `.`, `_`, and `-`, starting with a letter or digit, and no `..`. `claude plugin validate` fails any other name, because Claude Code [can't install plugins from a marketplace that… |
|
v2.1.295PreModelSwitch |
Documentation arrivedAnthropic's documentation has since written up PreModelSwitch, on Hooks reference.
the line in the docsA PreModelSwitch hook that doesn't respond before its timeout blocks the switch. For what a timeout does on other events, see [Timeouts](#timeouts). The default timeout for this event is 30 seconds. `PreModelSwitch` runs `command`, `http`,… |
|
v2.1.287asyncRewake |
Documentation arrivedAnthropic's documentation has since written up asyncRewake, on Hooks reference.
the line in the docs* **Background command hooks**: command hooks that set [`async` or `asyncRewake`](#run-hooks-in-the-background) |
|
v2.1.293PermissionRequest |
Documentation arrivedAnthropic's documentation has since written up PermissionRequest, on Hooks reference.
the line in the docsWith `"block"` set, a failure does what [exit code 2 does on that event](#exit-code-2-behavior-per-event), except on `PermissionRequest`, where it denies the request. For example, a `PreToolUse` failure blocks the tool call and a `UserProm… |
|
v2.1.295onFailure |
Documentation arrivedAnthropic's documentation has since written up onFailure, on Hooks reference.
the line in the docs| `onFailure` | no | What happens to the action when the hook fails: `"continue"`, the default, or `"block"`. See [Block the action when a hook fails](#block-the-action-when-a-hook-fails). Requires Claude Code v2.1.295 or later | |
|
v2.1.295onFailure |
Documentation arrivedAnthropic's documentation has since written up onFailure, on Hooks reference.
the line in the docs| `onFailure` | no | What happens to the action when the hook fails: `"continue"`, the default, or `"block"`. See [Block the action when a hook fails](#block-the-action-when-a-hook-fails). Requires Claude Code v2.1.295 or later | |
|
v2.1.290claude logs |
Documentation arrivedA small documentation edit on Manage multiple agents with agent view touched a line naming claude logs after this was published.
the line in the docs`claude attach` and `claude logs` can take part of a session's name in place of the ID, as in `claude logs "auth refactor"`. Passing a name requires Claude Code v2.1.290 or later. |
|
v2.1.290claude logs |
Documentation arrivedA small documentation edit on Manage multiple agents with agent view touched a line naming claude logs after this was published.
the line in the docs`claude attach` and `claude logs` can take part of a session's name in place of the ID, as in `claude logs "auth refactor"`. Passing a name requires Claude Code v2.1.290 or later. |
|
v2.1.290claude attach |
Documentation arrivedA small documentation edit on Manage multiple agents with agent view touched a line naming claude attach after this was published.
the line in the docs`claude attach` and `claude logs` can take part of a session's name in place of the ID, as in `claude logs "auth refactor"`. Passing a name requires Claude Code v2.1.290 or later. |
|
v2.1.285claude attach |
Documentation arrivedA small documentation edit on Manage multiple agents with agent view touched a line naming claude attach after this was published.
the line in the docs`claude attach` and `claude logs` can take part of a session's name in place of the ID, as in `claude logs "auth refactor"`. Passing a name requires Claude Code v2.1.290 or later. |
|
v2.1.285claude attach |
Documentation arrivedA small documentation edit on Manage multiple agents with agent view touched a line naming claude attach after this was published.
the line in the docs`claude attach` and `claude logs` can take part of a session's name in place of the ID, as in `claude logs "auth refactor"`. Passing a name requires Claude Code v2.1.290 or later. |
|
v2.1.290claude plugin list |
Documentation arrivedAnthropic's documentation has since written up claude plugin list, on Troubleshoot plugins.
the line in the docsOn Windows, you run `claude plugin uninstall` at project or local scope and it reports success, but `claude plugin list` or `/plugin` still lists the plugin. |
|
v2.1.289claude plugin list |
Documentation arrivedAnthropic's documentation has since written up claude plugin list, on Troubleshoot plugins.
the line in the docsOn Windows, you run `claude plugin uninstall` at project or local scope and it reports success, but `claude plugin list` or `/plugin` still lists the plugin. |
|
v2.1.293claude-haiku-5-5 |
Documentation arrivedA small documentation edit on overview touched a line naming claude-haiku-5-5 after this was published.
the line in the docs* Code written for Claude Haiku 4.5 can break on Claude Haiku 5.5. Manual extended thinking (`budget_tokens`), the sampling parameters `temperature`, `top_p`, and `top_k`, and assistant message prefill can each return a 400 error. `compute… |
|
v2.1.293claude-haiku-5-5 |
Documentation arrivedA small documentation edit on overview touched a line naming claude-haiku-5-5 after this was published.
the line in the docs* Code written for Claude Haiku 4.5 can break on Claude Haiku 5.5. Manual extended thinking (`budget_tokens`), the sampling parameters `temperature`, `top_p`, and `top_k`, and assistant message prefill can each return a 400 error. `compute… |
|
v2.1.293claude-haiku-5-5 |
Documentation arrivedA small documentation edit on overview touched a line naming claude-haiku-5-5 after this was published.
the line in the docs* Code written for Claude Haiku 4.5 can break on Claude Haiku 5.5. Manual extended thinking (`budget_tokens`), the sampling parameters `temperature`, `top_p`, and `top_k`, and assistant message prefill can each return a 400 error. `compute… |
|
v2.1.293claude-haiku-5-5 |
Documentation arrivedA small documentation edit on overview touched a line naming claude-haiku-5-5 after this was published.
the line in the docs* Code written for Claude Haiku 4.5 can break on Claude Haiku 5.5. Manual extended thinking (`budget_tokens`), the sampling parameters `temperature`, `top_p`, and `top_k`, and assistant message prefill can each return a 400 error. `compute… |
|
v2.1.293haiku |
Documentation arrivedA small documentation edit on overview touched a line naming haiku after this was published.
the line in the docs* Code written for Claude Haiku 4.5 can break on Claude Haiku 5.5. Manual extended thinking (`budget_tokens`), the sampling parameters `temperature`, `top_p`, and `top_k`, and assistant message prefill can each return a 400 error. `compute… |
|
v2.1.288output_config.format |
Documentation arrivedAnthropic's documentation has since written up output_config.format, on migration-guide.
the line in the docs11. If you use [structured outputs](https://platform.claude.com/docs/en/build-with-claude/structured-outputs) (`output_config.format` or `strict: true` tools) on Amazon Bedrock, describe the format in the prompt or use a tool without `stri… |
|
v2.1.293SessionEnd |
Documentation arrivedAnthropic's documentation has since written up SessionEnd, on events-and-streaming.
the line in the docs&& idle.StopReason?.Value is BetaManagedAgentsSessionEndTurn) |
|
v2.1.271multiplier |
Documentation arrivedA small documentation edit on data-residency touched a line naming multiplier after this was published.
the line in the docs* **Claude 4.6 and later models:** US-only inference (`inference_geo: "us"`) is priced at 1.1x the standard rate across all token pricing categories (input tokens, output tokens, cache writes, and cache reads). On Claude Haiku 5.5, the mul… |
|
v2.1.275maxTokens |
Documentation arrivedA small documentation edit on java touched a line naming maxTokens after this was published.
the line in the docsAvoid setting a large `maxTokens` value without using streaming. Some networks may drop idle connections after a certain period of time, which can cause the request to fail or [time out](https://platform.claude.com/docs/en/cli-sdks-librari… |
|
v2.1.293/compact |
Documentation arrivedAnthropic's documentation has since written up /compact, on thinking-troubleshooting.
the line in the docsTo fix it, keep the history append-only: pass earlier turns back exactly as sent and received, add instructions with a [mid-conversation system message](https://platform.claude.com/docs/en/build-with-claude/mid-conversation-system-messages… |
|
v2.1.288requestTimeout |
Documentation arrivedAnthropic's documentation has since written up requestTimeout, on mid-conversation-effort-example.
the line in the docsctx, cancel := context.WithTimeout(ctx, requestTimeoutSeconds*time.Second) |
|
v2.1.288requestTimeout |
Documentation arrivedAnthropic's documentation has since written up requestTimeout, on mid-conversation-effort-example.
the line in the docsctx, cancel := context.WithTimeout(ctx, requestTimeoutSeconds*time.Second) |
|
v2.1.293claude-haiku-5-5 |
Documentation arrivedAnthropic's documentation has since written up claude-haiku-5-5, on effort.
the line in the docsClaude Haiku 5.5 supports all five effort levels, and `medium` is the default. Effort is the main control for how much the model thinks, and with it quality, latency, and cost. **Start with `medium`** for most work, including agentic codin… |
|
v2.1.293claude-haiku-5-5 |
Documentation arrivedAnthropic's documentation has since written up claude-haiku-5-5, on effort.
the line in the docsClaude Haiku 5.5 supports all five effort levels, and `medium` is the default. Effort is the main control for how much the model thinks, and with it quality, latency, and cost. **Start with `medium`** for most work, including agentic codin… |
|
v2.1.293claude-haiku-5-5 |
Documentation arrivedAnthropic's documentation has since written up claude-haiku-5-5, on effort.
the line in the docsClaude Haiku 5.5 supports all five effort levels, and `medium` is the default. Effort is the main control for how much the model thinks, and with it quality, latency, and cost. **Start with `medium`** for most work, including agentic codin… |
|
v2.1.293claude-haiku-5-5 |
Documentation arrivedAnthropic's documentation has since written up claude-haiku-5-5, on effort.
the line in the docsClaude Haiku 5.5 supports all five effort levels, and `medium` is the default. Effort is the main control for how much the model thinks, and with it quality, latency, and cost. **Start with `medium`** for most work, including agentic codin… |
|
v2.1.293/compact |
Documentation arrivedA small documentation edit on compaction-on-demand touched a line naming /compact after this was published.
the line in the docs* **Mid-conversation system messages and tool changes.** `role: "system"` messages inside the summarized range are summarized too, so their text instructions stop applying once the block replaces them. A [per-message effort](https://platfo… |
|
v2.1.288claude update |
Documentation arrivedA small documentation edit on claude-in-amazon-bedrock touched a line naming claude update after this was published.
the line in the docsUse Claude Code 2.1.255 or later with Claude Fable 5.1 on Amazon Bedrock, and 2.1.280 or later with Claude Opus 5.5 and Claude Haiku 5.5; run `claude update` to upgrade. |
|
v2.1.290PreToolUse |
Documentation arrivedA small documentation edit on Code in Claude Desktop on 3P touched a line naming PreToolUse after this was published.
the line in the docs| `builtinToolPolicy` | Tools set to `"ask"` require approval on each call in Code sessions, enforced via a PreToolUse hook and Claude Code `permissions.ask` rules. Argument-scoped rules such as `Bash(curl *)` are [applied as managed polic… |
|
v2.1.289PreToolUse |
Documentation arrivedA small documentation edit on Code in Claude Desktop on 3P touched a line naming PreToolUse after this was published.
the line in the docs| `builtinToolPolicy` | Tools set to `"ask"` require approval on each call in Code sessions, enforced via a PreToolUse hook and Claude Code `permissions.ask` rules. Argument-scoped rules such as `Bash(curl *)` are [applied as managed polic… |
|
v2.1.282permissions.ask |
Documentation arrivedA small documentation edit on Code in Claude Desktop on 3P touched a line naming permissions.ask after this was published.
the line in the docs| `builtinToolPolicy` | Tools set to `"ask"` require approval on each call in Code sessions, enforced via a PreToolUse hook and Claude Code `permissions.ask` rules. Argument-scoped rules such as `Bash(curl *)` are [applied as managed polic… |
|
v2.1.281builtinToolPolicy |
Documentation arrivedA small documentation edit on Code in Claude Desktop on 3P touched a line naming builtinToolPolicy after this was published.
the line in the docs| `builtinToolPolicy` | Tools set to `"ask"` require approval on each call in Code sessions, enforced via a PreToolUse hook and Claude Code `permissions.ask` rules. Argument-scoped rules such as `Bash(curl *)` are [applied as managed polic… |
|
v2.1.281disallowedTools |
Documentation arrivedA small documentation edit on Create custom subagents touched a line naming disallowedTools after this was published.
the line in the docsThe full content of each listed skill is injected into the subagent's context at startup, up to the first 32 distinct names in the list. This field controls which skills are preloaded, not which skills the subagent can access: without it, … |
|
v2.1.281disallowedTools |
Documentation arrivedA small documentation edit on Create custom subagents touched a line naming disallowedTools after this was published.
the line in the docsThe full content of each listed skill is injected into the subagent's context at startup, up to the first 32 distinct names in the list. This field controls which skills are preloaded, not which skills the subagent can access: without it, … |
|
v2.1.281disallowedTools |
Documentation arrivedA small documentation edit on Create custom subagents touched a line naming disallowedTools after this was published.
the line in the docsThe full content of each listed skill is injected into the subagent's context at startup, up to the first 32 distinct names in the list. This field controls which skills are preloaded, not which skills the subagent can access: without it, … |
|
v2.1.281disallowedTools |
Documentation arrivedA small documentation edit on Create custom subagents touched a line naming disallowedTools after this was published.
the line in the docsThe full content of each listed skill is injected into the subagent's context at startup, up to the first 32 distinct names in the list. This field controls which skills are preloaded, not which skills the subagent can access: without it, … |
|
v2.1.290claude attach |
Documentation arrivedAnthropic's documentation has since written up claude attach, on Manage multiple agents with agent view.
the line in the docs* `claude attach` prints `This session has no saved transcript`. |
|
v2.1.285API_TIMEOUT_MS |
Documentation arrivedA small documentation edit on Agent SDK reference - Python touched a line naming API_TIMEOUT_MS after this was published.
the line in the docs* `CLAUDE_CODE_MAX_RETRIES`: maximum API retries. Default `10`, capped at `15`. Each retry gets its own `API_TIMEOUT_MS` window. |
|
v2.1.293managedMcpServers |
Documentation arrivedAnthropic's documentation has since written up managedMcpServers, on Changelog.
the line in the docs* Fixed tool calls to an OAuth-authenticated server in `managedMcpServers` failing until the next scheduled token refresh when the server rejected the token early, sometimes with a blank error, and failing for up to 30 minutes after the de… |