What this read moved
1–10 of 10monitoring-usage Changed · +5 / -0 lines
from line 1113
11131113* `num_non_blocking_error`: Count that failed without blocking
11141114* `num_cancelled`: Count cancelled before completion
11151115* `total_duration_ms`: Wall-clock duration of all matching hooks
1116* `stdout_chars`: Total characters of stdout across the matching hooks that succeeded. Requires Claude Code v2.1.280 or later
1117* `additional_context_chars`: Total characters of `additionalContext` returned by the matching hooks. Requires Claude Code v2.1.280 or later
1118* `system_message_chars`: Total characters of `systemMessage` returned by the matching hooks. Requires Claude Code v2.1.280 or later
1119* `initial_user_message_chars`: Total characters of `initialUserMessage` returned by the matching hooks. Requires Claude Code v2.1.280 or later
1120* `num_outputs_persisted`: Number of hook outputs over the [10,000-character cap](/docs/en/hooks#json-output) that Claude Code saved to a file. Requires Claude Code v2.1.280 or later
11161121* `managed_only`: `"true"` when only managed-policy hooks are permitted
11171122* `hook_source`: `"policySettings"` or `"merged"`
11181123* `safe_mode`: `"true"` when the session was started with [`--safe-mode`](/docs/en/cli-reference), `"false"` otherwise. Requires Claude Code v2.1.169 or later
plugin-marketplaces Changed · +12 / -28 lines
from line 783
783783
784784#### Background auto-updates
785785
786By default, the background refresh disables git credential helpers when it checks the marketplace's remote for new commits, so the check can't authenticate to private repositories over HTTPS even when a helper is configured. SSH remotes aren't affected: a key loaded in `ssh-agent` authenticates the background check the same way as the commands you run.
786The background refresh checks the marketplace's remote for new commits with your configured git credential helpers, the same as the commands you run. For SSH remotes, a key loaded in `ssh-agent` authenticates the check. Claude Code runs the check non-interactively: it turns off git's terminal prompts and askpass programs, and tells credential helpers not to prompt. Whether the check can authenticate to a private repository over HTTPS depends on your helper:
787787
788When the check finds new commits, or fails because it can't reach or authenticate to the remote, Claude Code clones the marketplace again and swaps the new clone in. If that clone fails, the existing checkout stays in place. The re-clone does use your stored git credentials, but it can [time out on large repositories](#git-operations-time-out), so private-marketplace auto-updates may fail intermittently.
788* A helper that can supply a stored credential without prompting authenticates the check. Git Credential Manager, the macOS Keychain helper, and `git-credential-store` work this way once they hold a credential for the host.
789* A helper that needs to prompt you can't answer in the background. The update fails quietly and the existing checkout stays in place, so your plugins keep working from the last synced state. Run `/plugin marketplace update <name>` to refresh the marketplace with your credentials.
789790
791When the check finds the checkout up to date, Claude Code leaves it as it is. When the check finds new commits, or fails because it can't reach or authenticate to the remote, Claude Code clones the marketplace again and swaps the new clone in. If that clone fails, the existing checkout stays in place. The re-clone can [time out on large repositories](#git-operations-time-out).
792
790793Two settings make private marketplaces behave predictably:
791794
792795* Set `CLAUDE_CODE_PLUGIN_KEEP_MARKETPLACE_ON_FAILURE=1` to keep the existing checkout without attempting the re-clone when the background check can't reach or authenticate to the remote. Your plugins keep working from the last synced state, and manual updates with `/plugin marketplace update` still authenticate with your credentials.
793* Configure a git credential helper, for example with `gh auth setup-git` for GitHub, so the re-clone can authenticate without prompting.
796* Configure a git credential helper, for example with `gh auth setup-git` for GitHub, so the background check and the re-clone can authenticate without prompting.
794797
795798Setting a provider token such as `GITHUB_TOKEN` in your environment doesn't by itself enable background authentication. Tokens take effect only through a configured credential helper, for example the `gh` CLI's helper, which reads `GH_TOKEN` and `GITHUB_TOKEN`.
796799
797To make the background check itself authenticate over HTTPS, configure a global git URL rewrite. The rewrite embeds a token in the remote URL, so it takes effect even though the background check disables credential helpers. When the check finds the checkout up to date, Claude Code skips the re-clone. The following example rewrites the marketplace repository's URL to include an access token:
798
799```bash theme={null}
800git config --global url."https://x-access-token:[email protected]/acme-corp/plugins".insteadOf "https://github.com/acme-corp/plugins"
801```
802
803Scope the rewrite to the marketplace repository or organization path. A rewrite whose base is only the host applies to every fetch and push to that host on the machine and overrides your normal credentials, including pushes to your own repositories.
804
805Each provider expects a different username in the rewritten URL, and the same path scoping applies to every provider. For self-hosted servers, replace the hostname with your server's hostname:
806
807| Provider | Rewritten URL form |
808| :-------- | :---------------------------------------------------------------- |
809| GitHub | `https://x-access-token:[email protected]/acme-corp/plugins` |
810| GitLab | `https://oauth2:[email protected]/acme-corp/plugins` |
811| Bitbucket | `https://x-token-auth:[email protected]/acme-corp/plugins` |
812
813The rewrite stores the token in plaintext in your gitconfig, so use a token with read-only access to the marketplace repository.
814
815800<Note>
816801 In CI/CD environments, configure a git credential helper before installing plugins from private repositories. On GitHub Actions, export a token with read access to the marketplace repository as `GH_TOKEN`, then run `gh auth setup-git`. The default workflow token can only access the workflow's own repository, so a private marketplace in another repository needs a personal access token or app token.
817
818 If you configure a global URL rewrite in the pipeline, the rewrite also authenticates the background check directly.
819802</Note>
820803
821804### Distribute through organization settings
from line 1467
14841467
14851468For background auto-updates:
14861469
1487* By default, background refreshes disable git credential helpers when they check the remote for new commits, so the check can't authenticate over HTTPS. SSH remotes with a key loaded in `ssh-agent` still authenticate
1488* When the check can't authenticate, Claude Code re-clones the marketplace with your stored credentials, but the re-clone may time out on large repositories
1470* The background check uses your configured git credential helpers but never prompts, so your helper must be able to answer with a stored credential. SSH remotes with a key loaded in `ssh-agent` also authenticate
1471* If your helper needs to prompt you, the background update fails quietly and the existing checkout stays in place. Sign in to your helper first so it holds a credential for the host. For GitHub, run `gh auth login`, then `gh auth setup-git`
1472* When the check finds new commits, or can't reach or authenticate to the remote, Claude Code re-clones the marketplace with the same credentials. The re-clone may time out on large repositories
14891473* Set `CLAUDE_CODE_PLUGIN_KEEP_MARKETPLACE_ON_FAILURE=1` to keep the existing checkout without attempting the re-clone when the background check can't reach or authenticate to the remote
1490* Configure a git credential helper, for example `gh auth setup-git`, so the re-clone can authenticate
14911474* If the re-clone times out on a large repository, increase the limit with [`CLAUDE_CODE_PLUGIN_GIT_TIMEOUT_MS`](#git-operations-time-out)
1492* Configure a [git URL rewrite](#private-repositories) scoped to the marketplace repository so the background check authenticates directly
14931475* Or update private marketplaces manually with `/plugin marketplace update <name>`, which uses your credentials
1476
1477Before v2.1.280, the background check ran without your credential helpers and couldn't authenticate to private repositories over HTTPS.
14941478
14951479### Marketplace updates fail in offline environments
14961480
skills Changed · +8 / -0 lines
### Personal skills disappeared
from line 1084
10841084
10851085To raise the budget, set the [`skillListingBudgetFraction`](/docs/en/settings-reference#skilllistingbudgetfraction) setting (for example, `0.02` = 2%) or the `SLASH_COMMAND_TOOL_CHAR_BUDGET` environment variable to a fixed character count. To free budget for other skills, set low-priority entries to `"name-only"` in [`skillOverrides`](#override-skill-visibility-from-settings) so they list without a description. You can also trim the `description` and `when_to_use` text at the source: put the key use case first, since each entry's combined text is capped at 1,536 characters regardless of budget. The cap is configurable with [`skillListingMaxDescChars`](/docs/en/settings-reference#skilllistingmaxdescchars).
10861086
1087### Personal skills disappeared
1088
1089If skill folders you created in `~/.claude/skills/` are gone, look in `~/.claude/skills/.trash/`. When Claude Code [syncs skills from claude.ai](#how-synced-skills-behave), it downloads them into the separate `synced` subfolder and doesn't move or delete the folders you create.
1090
1091Before v2.1.280, a file named `manifest.json` in `~/.claude/skills/` caused Claude Code to move the skill folders that file listed into a timestamped folder under `~/.claude/skills/.trash/`, and those skills stopped loading.
1092
1093To restore a skill, move its folder from the timestamped folder back into `~/.claude/skills/`. Do this before the [retention sweep](/docs/en/claude-directory#cleaned-up-automatically) deletes trash entries, by default 30 days after they were moved to the trash.
1094
10871095## Related resources
10881096
10891097* **[Debug your configuration](/docs/en/debug-your-config)**: diagnose why a skill isn't appearing or triggering
ultrareview Changed · +8 / -2 lines
from line 137
137137
138138Use `/tasks` to see running and completed reviews, open the detail view for a review, or stop a review that is in progress. If you stop a review, Claude Code archives the cloud session and doesn't return partial findings.
139139
140Claude can also tell you that a review was stopped or that its session wasn't found:
141
142* If the review's cloud session is stopped or [archived](/docs/en/claude-code-on-the-web#archive-sessions) on claude.ai before the review finishes, Claude tells you it was stopped.
143* If the review's cloud session was deleted, or you've signed in to a different Claude account or organization since launching it, Claude tells you the session wasn't found.
144* If you switched accounts, the review may still finish under the account that started it. If the review is still running, sign back in as that account and resume the conversation with `claude --resume` to re-attach it.
145
140146When the review finishes, Claude Code shows the verified findings as a notification in your session. Each finding includes the file location and an explanation of the issue so you can ask Claude to fix it directly.
141147
142148## Run ultrareview non-interactively
from line 175
169175The subcommand exits with one of three codes:
170176
171177* **0**: the review completed, with or without findings
172* **1**: the review failed to launch, the cloud session errored, or the timeout elapsed
178* **1**: the review failed to launch or was stopped before it finished, the cloud session errored, or the timeout elapsed
173179* **130**: you interrupted the subcommand with Ctrl-C
174180
175181If you interrupt the subcommand, the remote review keeps running; follow the session URL printed to stderr to watch it in the browser.
from line 182
176182
177183With `--post`, the subcommand starts the post right after printing the findings, and prints the link to stderr.
178184
179* If the run fails, times out, or you interrupt it, the subcommand posts nothing.
185* If the run fails, is stopped, or times out, or if you interrupt it, the subcommand posts nothing.
180186* If the review completes but the comment isn't posted, Claude Code prints the reason to stderr, and the findings stay on stdout so you can post them by hand.
181187
182188For automatic reviews on GitHub pull requests, [Code Review](/docs/en/code-review) integrates with your repository directly and posts findings as inline PR comments without a CLI step.
voice-dictation Changed · +10 / -0 lines
## Cancel a recording
from line 61
6161
6262The first couple of key-repeat characters type into the input during warmup and are removed automatically when recording activates. A single `Space` tap still types a space, since hold detection only triggers on rapid repeat.
6363
64Holding or tapping `Space` starts dictation only where the keypress would otherwise type into the prompt. In the [transcript viewer](/docs/en/interactive-mode#transcript-viewer), `Space` pages through the conversation, and in [vim mode](/docs/en/interactive-mode#vim-editor-mode) outside INSERT it is a command. A [rebound modifier combination](#rebind-the-dictation-key) like `meta+k` never types text, so it starts dictation from those places too.
65
6466<Tip>
6567 To skip the warmup, switch to [tap mode](#tap-to-record-and-send) with `/voice tap`, or [rebind to a modifier combination](#rebind-the-dictation-key) like `meta+k`. Modifier combos start recording on the first keypress.
6668</Tip>
from line 88
8688The three-word threshold counts words for languages written without spaces. Japanese, Chinese, and Thai transcripts count individual words, so they auto-submit in tap mode and in hold mode with `autoSubmit`.
8789
8890The first tap only starts recording when the prompt input is empty, so you can still type spaces normally while composing a message. The second tap stops recording regardless of input contents. Recording also stops automatically after 15 seconds of silence or two minutes total.
91
92## Cancel a recording
93
94Press `Esc` or `Ctrl+C` to cancel a dictation instead of finalizing it. Claude Code stops the microphone, discards the transcript, and restores the prompt to what it held before the recording started.
95
96Both keys also cancel while a finished recording's transcript is still processing. A prompt you edited or submitted during processing stays as you left it.
97
98Neither key does anything else in the press that cancels: `Esc` doesn't interrupt Claude's response, and `Ctrl+C` doesn't clear the prompt or count as the first of the [two presses that exit Claude Code](/docs/en/interactive-mode#general-controls).
8999
90100## Change the dictation language
91101
vs-code Changed · +22 / -1 lines
### Paste text
from line 101
101101 * **Auto**: a classifier reviews most actions instead of asking you. See [auto mode](/docs/en/permission-modes#eliminate-prompts-with-auto-mode) for what it reviews and blocks.
102102 * **Manual**: Claude asks permission before file edits and most shell commands.
103103 * **Plan**: Claude describes what it will do and waits for approval before making changes. VS Code automatically opens the plan as a full Markdown document where you can add inline comments to give feedback before Claude begins.
104
105 You can also type `/plan` in the prompt box. Requires Claude Code v2.1.280 or later.
106
107 * `/plan`: switches to plan mode. If you're already in plan mode, shows the current plan instead.
108 * `/plan` with a task, such as `/plan fix the auth bug`: switches to plan mode and starts planning that task.
109 * `/plan open`: when you're already in plan mode, opens the plan file in the editor.
104110 * **Edit automatically**: Claude makes edits without asking.
105111* **Model**: select **Switch model…** from the command menu to change the model mid-session. You can also click the model name at the bottom of the prompt box to open the same picker.
106112
from line 113
107113 When the current model supports [effort levels](/docs/en/model-config#adjust-effort-level), the picker also shows an **Effort** row and the model name button shows the selected level. When you pick a level other than `max`, Claude Code saves it for the current model as your default, under [`modelSettings`](/docs/en/settings-reference#modelsettings) in your user settings; `max` applies to the current session only. The model name button and the **Effort** row require Claude Code v2.1.257 or later.
108114* **Command menu**: click `/` or type `/` to open the command menu. Options include attaching files, switching models, and toggling extended thinking.
109115
110 The Customize section provides access to MCP servers, commands, output styles, hooks, memory, instructions, permissions, and plugins. Items with a terminal icon open in the integrated terminal.
116 The Customize section includes entries such as MCP servers, commands, output styles, hooks, memory, instructions, permissions, and plugins. Items with a terminal icon open in the integrated terminal.
111117
112118 * To browse commands such as `/usage` or [`/remote-control`](/docs/en/remote-control), select **Slash commands** in the Customize section. A dialog lists them with a filter box. Pick one to run it. Typing `/` in the prompt box still suggests commands inline. Requires Claude Code v2.1.257 or later.
119
120 Typing `/skills` also opens this dialog. Each [skill](/docs/en/skills) row shows its [visibility](/docs/en/skills#override-skill-visibility-from-settings), such as **On** or **Name only**. Click the visibility to change it, except on rows marked **locked**, such as plugin skills. The `/skills` shortcut and the visibility controls require Claude Code v2.1.280 or later.
113121 * Select **Output styles** in the Customize section to pick an [output style](/docs/en/output-styles), including your custom styles. Requires Claude Code v2.1.257 or later.
114122
115123 To create a custom style instead, select **Build a custom style** from the **Output styles** menu. Claude Code writes the [style file](/docs/en/output-styles#create-a-custom-output-style) for you at the project or user level. Requires Claude Code v2.1.261 or later.
from line 127
119127
120128 Click a saved memory to read it in the dialog, where you can edit the text, delete the memory, or open its file in the editor. Viewing, editing, and deleting a memory in the dialog require Claude Code v2.1.275 or later.
121129 * Select **Instructions** in the Customize section to edit the [CLAUDE.md files](/docs/en/memory#claude-md-files) Claude reads. Pick a file to open it in the editor. If the file doesn't exist yet, Claude Code creates it first. Requires Claude Code v2.1.274 or later.
130 * Select **Status** in the Customize section, or type `/status`, to check the session's Claude Code version, account, model, and MCP server details. Requires Claude Code v2.1.280 or later.
131 * Select **Sandbox** in the Customize section, or type `/sandbox`, to see whether Claude's Bash commands run [sandboxed](/docs/en/sandboxing). You can switch the sandbox mode and add [excluded commands](/docs/en/settings-reference#sandbox-excludedcommands) there. Requires Claude Code v2.1.280 or later.
132 * Select **Claude in Chrome** in the Customize section, or type `/chrome`, to check and manage the [Claude in Chrome](/docs/en/chrome) connection. Both require signing in with a claude.ai account. Requires Claude Code v2.1.280 or later.
133 * Select **Export conversation** in the Context section, or type `/export`, to copy the conversation as plain text or save it to a file. Add a file name, such as `/export notes.txt`, to skip the dialog and choose where to save the file. Requires Claude Code v2.1.280 or later.
122134 * The Settings section includes **Enable Remote Control for all sessions**, which sets [`remoteControlAtStartup`](/docs/en/settings-reference#remotecontrolatstartup) to control whether [new interactive sessions connect to Remote Control automatically](/docs/en/remote-control#enable-remote-control-for-all-sessions). Requires Claude Code v2.1.203 or later.
123135
124136 When you turn the toggle on or off in a VS Code window, the change applies to the sessions already open in that VS Code window, not only to sessions you start afterwards. If you turn it off, the open sessions disconnect. With Claude Code v2.1.261 or later, the change also reaches sessions open in your other VS Code windows.
from line 180
168180* To attach an image, paste it from your clipboard into the prompt box.
169181* To attach files, hold `Shift` while dragging them into the prompt box.
170182* To remove an attachment from context, click the X on it.
183
184### Paste text
185
186Text you paste stays visible in the prompt box, rather than collapsing to a placeholder as it does [in the terminal](/docs/en/terminal-config#paste-large-content). In sessions where Claude Code [marks pasted text](/docs/en/terminal-config#how-claude-treats-pasted-text), Claude still sees a large paste as text you pasted rather than typed.
187
188Claude Code also removes [invisible Unicode characters](/docs/en/interactive-mode#invisible-characters-in-prompts) from text you paste into the prompt box and from anything else you send:
189
190* If a notice such as `Removed 3 invisible characters from the pasted text` appears when you paste, the text went in without those characters.
191* If a notice about removed characters appears when you send, nothing was sent. The cleaned text is back in the prompt box. Send again to send the text as shown.
171192
172193### Resume past conversations
173194
artifacts Changed · +4 / -0 lines
from line 302
302302| Organization policy | Customer-managed encryption keys (CMEK), HIPAA, and [Zero Data Retention](/docs/en/zero-data-retention) are not enabled for the organization. |
303303| Surface | Claude Code CLI, or the Claude desktop app version 1.13576.0 or later. [Claude Tag](https://claude.com/docs/claude-tag/overview) sessions can also publish artifacts when both Claude Tag and artifacts are enabled for the organization. Off by default in [Agent SDK](/docs/en/agent-sdk/overview), GitHub Action, and MCP-server contexts, and when [`CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC`](/docs/en/env-vars) is set. |
304304
305Whether artifacts are allowed for your organization comes from your organization's policy, which Claude Code loads from `api.anthropic.com`. When Claude Code can't load the policy, artifacts are unavailable. When you ask for one, Claude says why.
306
307If a proxy, VPN, or web filter is involved, ask your IT admin to let `api.anthropic.com` through. Claude Code keeps retrying in the background, and artifacts become available once the policy loads and allows them.
308
305309## Disable artifacts
306310
307311To turn artifacts off for your own sessions regardless of your organization's setting, use any of:
env-vars Changed · +2 / -0 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 297
297297| `CLAUDE_CODE_IDE_SKIP_VALID_CHECK` | Set to `1` to skip validation of IDE lockfile entries during connection. Use when auto-connect fails to find your IDE despite it running |
298298| `CLAUDE_CODE_MAX_CONCURRENT_SUBAGENTS` | How many [subagents](/docs/en/sub-agents#concurrent-subagent-limit) can be running in one session before the Agent tool refuses to spawn another (default: 20). Accepts a positive whole number in plain digits; anything else is ignored, so the variable can adjust the cap but can't disable it. Requires Claude Code v2.1.217 or later |
299299| `CLAUDE_CODE_MAX_CONTEXT_TOKENS` | Override the context window size Claude Code assumes for the active model. As of v2.1.193, how it applies depends on how Claude Code resolves the model ID; see [Correct the window for a gateway or custom model ID](/docs/en/model-config#correct-the-window-for-a-gateway-or-custom-model-id). Use this when routing to a model through `ANTHROPIC_BASE_URL` whose context window does not match the built-in size for its name |
300| `CLAUDE_CODE_MAX_MCP_DESCRIPTION_LENGTH` | Maximum length in characters of each MCP tool description and each MCP server's instructions that Claude Code sends to the model (default: 2048). Claude Code [truncates longer text](/docs/en/mcp#for-mcp-server-authors). Accepts a positive whole number in plain digits. Anything else is ignored and the default applies. Requires Claude Code v2.1.280 or later |
300301| `CLAUDE_CODE_MAX_OUTPUT_TOKENS` | Set the maximum number of output tokens for most requests. Defaults and caps vary by model; see [max output tokens](https://platform.claude.com/docs/en/about-claude/models/overview#latest-models-comparison). Claude Code defaults to 32000 for model IDs it doesn't recognize, such as gateway-specific names, and lowers values above a model's cap to the cap. Increasing this value reduces the effective context window available before [auto-compaction](/docs/en/costs#reduce-token-usage) triggers |
301302| `CLAUDE_CODE_MAX_RETRIES` | Override the number of times to retry failed API requests (default: 10). Capped at 15 as of v2.1.186; as of v2.1.199, `CLAUDE_CODE_RETRY_WATCHDOG` raises the default and removes the cap. For unattended sessions that need to wait through longer outages, set `CLAUDE_CODE_RETRY_WATCHDOG` instead |
302303| `CLAUDE_CODE_MAX_SUBAGENTS_PER_SESSION` | Removed in v2.1.224 and now a no-op. Previously capped the total number of [subagents](/docs/en/sub-agents) Claude could spawn with the Agent tool in one session (default: 200); spawning past the cap failed with `Subagent spawn limit reached`. The [concurrent subagent limit](/docs/en/sub-agents#concurrent-subagent-limit) and the [depth limit](/docs/en/sub-agents#let-subagents-spawn-their-own-subagents) still apply |
from line 318
317318| `CLAUDE_CODE_OAUTH_REFRESH_TOKEN` | OAuth refresh token for Claude.ai authentication. When set, `claude auth login` exchanges this token directly instead of opening a browser. Requires `CLAUDE_CODE_OAUTH_SCOPES`. Useful for provisioning authentication in automated environments |
318319| `CLAUDE_CODE_OAUTH_SCOPES` | Space-separated OAuth scopes the refresh token was issued with, such as `"user:profile user:inference user:sessions:claude_code"`. Required when `CLAUDE_CODE_OAUTH_REFRESH_TOKEN` is set |
319320| `CLAUDE_CODE_OAUTH_TOKEN` | OAuth access token for claude.ai authentication. Alternative to `/login` for SDK and automated environments. Takes precedence over keychain-stored credentials. Generate one with [`claude setup-token`](/docs/en/authentication#generate-a-long-lived-token). Unless you run [`/login`](/docs/en/authentication#authentication-precedence), Claude Code uses the token you set for the whole session. To replace an expired token, generate a new one and restart |
320| `CLAUDE_CODE_OPUS_4_6_FAST_MODE_OVERRIDE` | Removed in v2.1.160 and now a no-op. Previously pinned [fast mode](/docs/en/fast-mode) to Claude Opus 4.6 instead of the current default. Opus 4.6 no longer supports fast mode |
321| `CLAUDE_CODE_OTEL_CONTENT_MAX_LENGTH` | Maximum length of content-bearing OpenTelemetry attributes (model responses, tool content, system prompts, raw API bodies), truncation marker included, in UTF-16 code units (default: 61440, i.e. 60 KB). Raise it only if your telemetry backend accepts attribute values larger than 64 KB, or lower it to cut telemetry volume. Requires Claude Code v2.1.214 or later. See [Monitoring](/docs/en/monitoring-usage)
321| `CLAUDE_CODE_OPUS_4_6_FAST_MODE_OVERRIDE` | Removed in v2.1.160 and now a no-op. Previously pinned [fast mode](/docs/en/fast-mode) to Claude Opus 4.6 instead of the current default. Opus 4.6 no longer supports fast mode
mcp Changed · +3 / -1 lines
from line 1416
14161416* When Claude should search for your tools
14171417* Key capabilities your server provides
14181418
1419Claude Code truncates tool descriptions and server instructions at 2KB each. Keep them concise to avoid truncation, and put critical details near the start.
1419Claude Code truncates each tool description and each server's instructions at 2,048 characters by default. Keep them concise, and put critical details near the start.
1420
1421To change the limit for every MCP server in your session, set [`CLAUDE_CODE_MAX_MCP_DESCRIPTION_LENGTH`](/docs/en/env-vars#variables) to a number of characters. This variable requires Claude Code v2.1.280 or later.
14201422
14211423### Configure tool search
14221424
plugins Changed · +2 / -0 lines
from line 310
310310 To test a plugin together with a plugin it depends on, see [Test a plugin and its dependency locally](/docs/en/plugin-dependencies#test-a-plugin-and-its-dependency-locally).
311311</Tip>
312312
313To load plugins in a session where you can't add the flag, list their absolute paths in the [`CLAUDE_CODE_PLUGIN_DIRS`](/docs/en/env-vars#variables) environment variable instead. Claude Code loads each path as it loads a `--plugin-dir` path. These plugins load in addition to any you pass with `--plugin-dir`. [Project and local settings can't set this variable](/docs/en/settings-reference#variables-claude-code-ignores-in-env). `CLAUDE_CODE_PLUGIN_DIRS` requires Claude Code v2.1.280 or later.
314
313315Trying the plugin with `--plugin-dir` tells you it can work. To find out how often Claude actually reaches for it and gets the right result, run it against a set of test prompts with [`claude plugin eval`](/docs/en/plugin-evals). Each prompt runs several times with and without the plugin loaded, so you can see what the plugin contributes and catch regressions when you change it or a new model ships.
314316
315317To load several plugins from one place, pass a folder that holds them, such as `--plugin-dir ./plugins`. Loading a folder of plugins requires Claude Code v2.1.265 or later. Claude Code reads the folder's top level to decide which plugins load, and in an interactive session it also watches the folder for later changes: