83| `Your account does not have access to Claude. Please login again or contact your administrator.` | [Authentication](#oauth-token-revoked-or-expired) |83| `Your account does not have access to Claude. Please login again or contact your administrator.` | [Authentication](#oauth-token-revoked-or-expired) |
84| `API Error: 401 Invalid authentication credentials` | [Authentication](#api-error-401-invalid-authentication-credentials) |84| `API Error: 401 Invalid authentication credentials` | [Authentication](#api-error-401-invalid-authentication-credentials) |
85| `Login expired · Please run /login` | [Authentication](#login-expired) |85| `Login expired · Please run /login` | [Authentication](#login-expired) |
86| `The browser can't return the sign-in to Claude Code automatically` | [Authentication](#couldnt-start-the-local-sign-in-listener) |
87| `Couldn't start the local listener that receives the browser sign-in. Try again.` | [Authentication](#couldnt-start-the-local-sign-in-listener) |
86| `Failed to start OAuth callback server` | [Authentication](#failed-to-start-oauth-callback-server) |88| `Failed to start OAuth callback server` | [Authentication](#failed-to-start-oauth-callback-server) |
87| `Claude login not accepted · Run /login, then try again` | [Authentication](#claude-login-not-accepted) |89| `Claude login not accepted · Run /login, then try again` | [Authentication](#claude-login-not-accepted) |
88| `Artifacts need a claude.ai login` | [Authentication](#artifacts-need-a-claude-ai-login) |90| `Artifacts need a claude.ai login` | [Authentication](#artifacts-need-a-claude-ai-login) |
317| `Failed to write to <name>'s inbox — nothing was sent` | [Tool errors](#failed-to-write-to-a-teammate-inbox) |319| `Failed to write to <name>'s inbox — nothing was sent` | [Tool errors](#failed-to-write-to-a-teammate-inbox) |
318| `Failed to write the plan approval request to the lead's inbox — plan not submitted` | [Tool errors](#failed-to-write-to-a-teammate-inbox) |320| `Failed to write the plan approval request to the lead's inbox — plan not submitted` | [Tool errors](#failed-to-write-to-a-teammate-inbox) |
319| `Its agent definition was not restored: the folder its definition file came from is not trusted` | [Tool errors](#teammate-agent-definition-not-restored) |321| `Its agent definition was not restored: the folder its definition file came from is not trusted` | [Tool errors](#teammate-agent-definition-not-restored) |
322| `Only a teammate can approve a shutdown request` | [Tool errors](#only-a-teammate-can-approve-a-shutdown-request) |
320| `Message too large for cross-session delivery` | [Tool errors](#message-too-large-for-cross-session-delivery) |323| `Message too large for cross-session delivery` | [Tool errors](#message-too-large-for-cross-session-delivery) |
321| `Too many messages to this session just now` | [Tool errors](#too-many-messages-to-this-session-just-now) |324| `Too many messages to this session just now` | [Tool errors](#too-many-messages-to-this-session-just-now) |
322| `Cross-session message was dropped at the recipient session's inbox` | [Tool errors](#cross-session-message-dropped-at-the-inbox) |325| `Cross-session message was dropped at the recipient session's inbox` | [Tool errors](#cross-session-message-dropped-at-the-inbox) |
332| `Your disk quota is full on the filesystem with Claude Code's temp directory <dir> (EDQUOT)` | [Tool errors](#disk-quota-or-temp-filesystem-is-full) |335| `Your disk quota is full on the filesystem with Claude Code's temp directory <dir> (EDQUOT)` | [Tool errors](#disk-quota-or-temp-filesystem-is-full) |
333| `The filesystem with Claude Code's temp directory <dir>, or your disk quota on it, is full (ENOSPC)` | [Tool errors](#disk-quota-or-temp-filesystem-is-full) |336| `The filesystem with Claude Code's temp directory <dir>, or your disk quota on it, is full (ENOSPC)` | [Tool errors](#disk-quota-or-temp-filesystem-is-full) |
334| `Command output was lost: the temp filesystem at <dir> is full` / `is out of inodes` | [Tool errors](#disk-quota-or-temp-filesystem-is-full) |337| `Command output was lost: the temp filesystem at <dir> is full` / `is out of inodes` | [Tool errors](#disk-quota-or-temp-filesystem-is-full) |
338| `The command was not started: this computer's disk is full (ENOSPC)` / `(EDQUOT)` | [Tool errors](#disk-quota-or-temp-filesystem-is-full) |
335| `File is not valid UTF-8. It may use a legacy encoding such as Windows-1252, Shift-JIS or GBK, or be binary` | [Tool errors](#file-is-not-valid-utf-8) |339| `File is not valid UTF-8. It may use a legacy encoding such as Windows-1252, Shift-JIS or GBK, or be binary` | [Tool errors](#file-is-not-valid-utf-8) |
336| `the source file is not valid UTF-8 text` / `the source file is not valid UTF-16 text` | [Tool errors](#the-source-file-is-not-valid-utf-8-text) |340| `the source file is not valid UTF-8 text` / `the source file is not valid UTF-16 text` | [Tool errors](#the-source-file-is-not-valid-utf-8-text) |
337| `the source file has the replacement character U+FFFD` | [Tool errors](#the-source-file-is-not-valid-utf-8-text) |341| `the source file has the replacement character U+FFFD` | [Tool errors](#the-source-file-is-not-valid-utf-8-text) |
353| `Can't open — this session is running in another terminal` | [Background session errors](#this-session-is-running-in-another-terminal) |357| `Can't open — this session is running in another terminal` | [Background session errors](#this-session-is-running-in-another-terminal) |
354| `This conversation is already open in another running Claude session` | [Background session errors](#this-session-is-running-in-another-terminal) |358| `This conversation is already open in another running Claude session` | [Background session errors](#this-session-is-running-in-another-terminal) |
355| `This session's saved conversation is no longer on disk` | [Background session errors](#this-sessions-saved-conversation-is-no-longer-on-disk) |359| `This session's saved conversation is no longer on disk` | [Background session errors](#this-sessions-saved-conversation-is-no-longer-on-disk) |
356| `Bypass permissions was requested at launch and ignored` | [Skip all checks with bypassPermissions mode](/docs/en/permission-modes#bypass-permissions-acceptance) |360| `Bypass permissions was requested at launch and ignored` | [Bypass permissions ignored in a background session](/docs/en/permission-modes#bypass-permissions-ignored-in-a-background-session) |
357| `kept <id> — its worktree is still at <path>` | [Background session errors](#worktree-has-commits-that-are-not-pushed-anywhere) |361| `kept <id> — its worktree is still at <path>` | [Background session errors](#worktree-has-commits-that-are-not-pushed-anywhere) |
358| `kept <id> — <n> unpushed commits on <branch>` | [Background session errors](#worktree-has-commits-that-are-not-pushed-anywhere) |362| `kept <id> — <n> unpushed commits on <branch>` | [Background session errors](#worktree-has-commits-that-are-not-pushed-anywhere) |
359| `kept <id> — worktree has commits that are not pushed anywhere` | [Background session errors](#worktree-has-commits-that-are-not-pushed-anywhere) |363| `kept <id> — worktree has commits that are not pushed anywhere` | [Background session errors](#worktree-has-commits-that-are-not-pushed-anywhere) |
404| `API Error: ANTHROPIC_FOUNDRY_RESOURCE must be a Foundry resource name` | [Configuration warnings](#anthropic-foundry-resource-must-be-a-foundry-resource-name) |408| `API Error: ANTHROPIC_FOUNDRY_RESOURCE must be a Foundry resource name` | [Configuration warnings](#anthropic-foundry-resource-must-be-a-foundry-resource-name) |
405| `headersHelper not run — this workspace has no persisted trust` | [Configuration warnings](#headershelper-not-run) |409| `headersHelper not run — this workspace has no persisted trust` | [Configuration warnings](#headershelper-not-run) |
406| `Invalid permission rule "..." was skipped: Malformed Tool(content) rule` | [Configuration warnings](#malformed-tool-content-rule) |410| `Invalid permission rule "..." was skipped: Malformed Tool(content) rule` | [Configuration warnings](#malformed-tool-content-rule) |
411| `matches no tool · check for typos` / `matches no known tool — check for typos` | [Configuration warnings](#matches-no-tool) |
407| `... is not matched by file permission checks` | [Configuration warnings](#is-not-matched-by-file-permission-checks) |412| `... is not matched by file permission checks` | [Configuration warnings](#is-not-matched-by-file-permission-checks) |
413| `Write(<path>) never matches · use Edit(<path>)` / `Glob(<path>) never matches · use Read(<path>)` | [Configuration warnings](#is-not-matched-by-file-permission-checks) |
408| `... has a wildcard before the rest of the command` | [Configuration warnings](#has-a-wildcard-before-the-rest-of-the-command) |414| `... has a wildcard before the rest of the command` | [Configuration warnings](#has-a-wildcard-before-the-rest-of-the-command) |
415| `allows any option at its * · use an exact value` | [Configuration warnings](#has-a-wildcard-before-the-rest-of-the-command) |
409| `Denying Bash also turns off the PowerShell tool, so Claude has neither` | [Configuration warnings](#denying-bash-also-turns-off-the-powershell-tool) |416| `Denying Bash also turns off the PowerShell tool, so Claude has neither` | [Configuration warnings](#denying-bash-also-turns-off-the-powershell-tool) |
410| `CLAUDE_CODE_DISABLE_1M_CONTEXT is set, but the 200K limit isn't enforced` | [Configuration warnings](#the-200k-limit-isnt-enforced) |417| `CLAUDE_CODE_DISABLE_1M_CONTEXT is set, but the 200K limit isn't enforced` | [Configuration warnings](#the-200k-limit-isnt-enforced) |
418| `CLAUDE_CODE_AUTO_COMPACT_WINDOW="<value>" reads as <N> tokens, so Claude Code treats it as the 100k minimum` | [Values and warnings for `CLAUDE_CODE_AUTO_COMPACT_WINDOW`](/docs/en/model-config#claude_code_auto_compact_window-values) |
419| `CLAUDE_CODE_AUTO_COMPACT_WINDOW="<value>" is too large, so Claude Code treats it as the 1M maximum` | [Values and warnings for `CLAUDE_CODE_AUTO_COMPACT_WINDOW`](/docs/en/model-config#claude_code_auto_compact_window-values) |
420| `CLAUDE_CODE_AUTO_COMPACT_WINDOW="<value>" isn't a usable size, so Claude Code ignores it` | [Values and warnings for `CLAUDE_CODE_AUTO_COMPACT_WINDOW`](/docs/en/model-config#claude_code_auto_compact_window-values) |
411| `Model "<model>" isn't recognized · context assumed to be 200k tokens` | [Correct the window for a gateway or custom model ID](/docs/en/model-config#correct-the-window-for-a-gateway-or-custom-model-id) |421| `Model "<model>" isn't recognized · context assumed to be 200k tokens` | [Correct the window for a gateway or custom model ID](/docs/en/model-config#correct-the-window-for-a-gateway-or-custom-model-id) |
412| `Claude Code doesn't recognize the model "<model>", so it assumes a context window of 200k tokens` | [Correct the window for a gateway or custom model ID](/docs/en/model-config#correct-the-window-for-a-gateway-or-custom-model-id) |422| `Claude Code doesn't recognize the model "<model>", so it assumes a context window of 200k tokens` | [Correct the window for a gateway or custom model ID](/docs/en/model-config#correct-the-window-for-a-gateway-or-custom-model-id) |
413| `[claude-code:unrecognized_model]` | [Configuration warnings](#unrecognized-model-id-on-a-request) |423| `[claude-code:unrecognized_model]` | [Configuration warnings](#unrecognized-model-id-on-a-request) |
501 511
502* Check [status.claude.com](https://status.claude.com), or the provider status page named in the message, for active incidents512* Check [status.claude.com](https://status.claude.com), or the provider status page named in the message, for active incidents
503* Wait a minute, then send your message again. Your original message is still in the conversation, so for a long prompt you can type `try again` instead of pasting the whole thing.513* Wait a minute, then send your message again. Your original message is still in the conversation, so for a long prompt you can type `try again` instead of pasting the whole thing.
514* If [`ANTHROPIC_BASE_URL`](/docs/en/env-vars) points at an [LLM gateway](/docs/en/llm-gateway) and every request fails this way, check whether your gateway is rejecting the tool search fields Claude Code sends by default. To check, set `ENABLE_TOOL_SEARCH=false` as [Tool search through an LLM gateway](/docs/en/mcp#tool-search-through-a-gateway) describes, then try again. Before v2.1.304, tool search was off behind a gateway unless you turned it on with `ENABLE_TOOL_SEARCH`.
504* If the error persists with no posted incident, run `/feedback` so Anthropic can investigate with your request details. See [Report an error](#report-an-error) if `/feedback` is unavailable in your environment.515* If the error persists with no posted incident, run `/feedback` so Anthropic can investigate with your request details. See [Report an error](#report-an-error) if `/feedback` is unavailable in your environment.
505 516
506### API Error: Repeated 529 Overloaded errors517### API Error: Repeated 529 Overloaded errors
1281* If `/status` shows only your login, run `/login` once. If the credential was revoked, a fresh login replaces it.1292* If `/status` shows only your login, run `/login` once. If the credential was revoked, a fresh login replaces it.
1282* If the same message returns for the same login account, the account or organization is no longer active. Check the account and organization that `/status` reports, and ask your organization admin to restore access.1293* If the same message returns for the same login account, the account or organization is no longer active. Check the account and organization that `/status` reports, and ask your organization admin to restore access.
1283* If [`ANTHROPIC_BASE_URL`](/docs/en/env-vars) points at an [LLM gateway](/docs/en/llm-gateway), the text after `401` is your gateway's message rather than Anthropic's, and `/login` doesn't change it. Fix the credential your gateway expects instead.1294* If [`ANTHROPIC_BASE_URL`](/docs/en/env-vars) points at an [LLM gateway](/docs/en/llm-gateway), the text after `401` is your gateway's message rather than Anthropic's, and `/login` doesn't change it. Fix the credential your gateway expects instead.
1295* If every request fails with a `401` once your gateway has the credential it expects, check whether the gateway is rejecting the tool search fields Claude Code sends by default. To check, set `ENABLE_TOOL_SEARCH=false` as [Tool search through an LLM gateway](/docs/en/mcp#tool-search-through-a-gateway) describes, then try again. Before v2.1.304, tool search was off behind a gateway unless you turned it on with `ENABLE_TOOL_SEARCH`.
1284 1296
1285### Login expired1297### Login expired
1286 1298
1352* On other platforms, run `/login` again1364* On other platforms, run `/login` again
1353* If the login still doesn't save, see [Not logged in or token expired](/docs/en/troubleshoot-install#not-logged-in-or-token-expired) for the keychain unlock command and other credential-storage recovery steps1365* If the login still doesn't save, see [Not logged in or token expired](/docs/en/troubleshoot-install#not-logged-in-or-token-expired) for the keychain unlock command and other credential-storage recovery steps
1354 1366
1355### Failed to start OAuth callback server1367### Couldn't start the local listener that receives the browser sign-in
1356 1368
1357When `/login`, `claude auth login`, or `claude setup-token` signs you in through the browser, Claude Code opens a listening port on `127.0.0.1` so your browser can return the sign-in result to it. This message means Claude Code couldn't open that port, and the sign-in stops before a browser window or login URL appears:1369Each of these messages from `/login`, `claude auth login`, or `claude setup-token` means Claude Code couldn't open the port on `127.0.0.1` where your browser returns the sign-in result.
1358 1370
1359```text theme={null}1371With the notice that begins `The browser can't return the sign-in to Claude Code automatically`, the sign-in continues without that port. No browser opens, and the notice appears above a URL where you sign in instead:
1360Failed to start OAuth callback server: Failed to start server. Is port 0 in use?1372
1373```text wrap theme={null}
1374The browser can't return the sign-in to Claude Code automatically. After you sign in, paste the code shown.
1361```1375```
1362 1376
1363If your message ends with `Is port 0 in use?`, the attempt to listen on the IPv4 loopback address `127.0.0.1` failed outright. Because the failure happens before a login URL exists, the `Paste code here if prompted` flow isn't available as a workaround.1377With the message that includes `Couldn't start the local listener`, the sign-in stopped. That happens, for example, when you pipe input into `claude auth login`:
1378
1379```text theme={null}
1380Couldn't start the local listener that receives the browser sign-in. Try again.
1381```
1364 1382
1365**What to do:**1383**What to do:**
1366 1384
1367* To sign in right away without the local listener: if you use a claude.ai subscription, run [`claude setup-token`](/docs/en/authentication#generate-a-long-lived-token) on a machine where sign-in works and set the token it prints as `CLAUDE_CODE_OAUTH_TOKEN` on this machine. Otherwise set `ANTHROPIC_API_KEY` to a key from the [Claude Console](https://platform.claude.com/settings/keys). [Authentication precedence](/docs/en/authentication#authentication-precedence) explains how Claude Code chooses between credentials.1385* **`The browser can't return the sign-in to Claude Code automatically`**: you don't need to fix anything first. Open the URL shown under the notice in a browser and sign in. The browser then shows a code: paste it into Claude Code at the `Paste code here >` prompt, and the sign-in continues from there.
1368* To use browser sign-in on this machine instead, Claude Code must be able to listen on `127.0.0.1`. If it runs inside a sandbox, check that the sandbox's policy allows listening on local ports, then run `/login` again. If it should be able to and still fails, run `/feedback` so the report includes your environment details.1386* **`Couldn't start the local listener that receives the browser sign-in`**: run the same command again. If it stops the same way, sign in with a token or API key instead of the browser. Or, if you piped input into `claude auth login`, run it in a terminal so it can take a pasted code.
1387* **To sign in with a token or API key instead of the browser**: if you use a claude.ai subscription, run [`claude setup-token`](/docs/en/authentication#generate-a-long-lived-token) on a machine where sign-in works. Then set the token it prints as the [`CLAUDE_CODE_OAUTH_TOKEN`](/docs/en/env-vars) environment variable on this machine. Otherwise set the `ANTHROPIC_API_KEY` environment variable to a key from the [Claude Console](https://platform.claude.com/settings/keys). [Authentication precedence](/docs/en/authentication#authentication-precedence) explains how Claude Code chooses between credentials.
1388* **To get the automatic browser return back**: Claude Code must be able to listen on `127.0.0.1`. If it runs inside a sandbox, check that the sandbox's policy allows listening on local ports, then sign in again. If listening should work there and the notice still appears, run `/feedback` so the report includes your environment details.
1389
1390Before v2.1.305, the sign-in stopped as soon as the port couldn't be opened and showed no URL to sign in at. The message read, for example, `Failed to start OAuth callback server: Failed to start server. Is port 0 in use?`. On those versions, update Claude Code and sign in again, or sign in with a token or API key instead of the browser.
1369 1391
1370### Claude login not accepted1392### Claude login not accepted
1371 1393
1836* Confirm you can reach the API host from the same shell by running `curl -I https://api.anthropic.com`. On Windows PowerShell use `curl.exe -I https://api.anthropic.com` so the built-in `Invoke-WebRequest` alias is not used.1858* Confirm you can reach the API host from the same shell by running `curl -I https://api.anthropic.com`. On Windows PowerShell use `curl.exe -I https://api.anthropic.com` so the built-in `Invoke-WebRequest` alias is not used.
1837* If you are behind a corporate proxy, set `HTTPS_PROXY` before launching Claude Code and see [Network configuration](/docs/en/network-config)1859* If you are behind a corporate proxy, set `HTTPS_PROXY` before launching Claude Code and see [Network configuration](/docs/en/network-config)
1838* If you route through an LLM gateway or relay, set [`ANTHROPIC_BASE_URL`](/docs/en/env-vars) to its address. See [Connect Claude Code to an LLM gateway](/docs/en/llm-gateway-connect) for setup.1860* If you route through an LLM gateway or relay, set [`ANTHROPIC_BASE_URL`](/docs/en/env-vars) to its address. See [Connect Claude Code to an LLM gateway](/docs/en/llm-gateway-connect) for setup.
1861* If `ANTHROPIC_BASE_URL` already points at your gateway and every request ends in `Connection dropped`, check whether your gateway is rejecting the tool search fields Claude Code sends by default. To check, set `ENABLE_TOOL_SEARCH=false` as [Tool search through an LLM gateway](/docs/en/mcp#tool-search-through-a-gateway) describes, then try again. Before v2.1.304, tool search was off behind a gateway unless you turned it on with `ENABLE_TOOL_SEARCH`.
1839* Ensure your firewall allows the hosts listed in [Network access requirements](/docs/en/network-config#network-access-requirements)1862* Ensure your firewall allows the hosts listed in [Network access requirements](/docs/en/network-config#network-access-requirements)
1840* Intermittent failures are [retried automatically](#automatic-retries); persistent failures point to a local network issue1863* Intermittent failures are [retried automatically](#automatic-retries)
1841 1864
1842If `curl` succeeds but Claude Code still fails, the cause is usually something between the runtime and the network rather than the network itself:1865If `curl` succeeds but Claude Code still fails, the cause is usually something between the runtime and the network rather than the network itself:
1843 1866
2190* Trim large `CLAUDE.md` memory files, or move instructions into [path-scoped rules](/docs/en/memory#path-specific-rules) that load only when relevant2213* Trim large `CLAUDE.md` memory files, or move instructions into [path-scoped rules](/docs/en/memory#path-specific-rules) that load only when relevant
2191* Auto-compact is on by default and normally prevents this error. If you turned it off in `/config` or with [`DISABLE_AUTO_COMPACT`](/docs/en/env-vars), turn it back on. If you keep it off, run `/compact` yourself before the window fills.2214* Auto-compact is on by default and normally prevents this error. If you turned it off in `/config` or with [`DISABLE_AUTO_COMPACT`](/docs/en/env-vars), turn it back on. If you keep it off, run `/compact` yourself before the window fills.
2192* If the error appears near 200K tokens on a model that accepts more, one cause is a model ID that Claude Code doesn't recognize. Claude Code assumes a 200K window for such an ID, so to keep working past that point, [give it the model's real context window](/docs/en/model-config#correct-the-window-for-a-gateway-or-custom-model-id).2215* If the error appears near 200K tokens on a model that accepts more, one cause is a model ID that Claude Code doesn't recognize. Claude Code assumes a 200K window for such an ID, so to keep working past that point, [give it the model's real context window](/docs/en/model-config#correct-the-window-for-a-gateway-or-custom-model-id).
2216* If the error appeared right after you stopped Claude while it was compacting the conversation, your conversation is unchanged and still large enough to need compacting. Run `/compact` and send your message again. To keep the error from recurring, [update Claude Code](/docs/en/setup#update-claude-code) to v2.1.304 or later. Before v2.1.304, stopping at that moment could end the turn with this error.
2193 2217
2194See [Explore the context window](/docs/en/context-window) for an interactive view of how context fills up.2218See [Explore the context window](/docs/en/context-window) for an interactive view of how context fills up.
2195 2219
2322 2346
2323### Extra inputs are not permitted2347### Extra inputs are not permitted
2324 2348
2325A proxy or LLM gateway between Claude Code and the API stripped the `anthropic-beta` request header, so the API rejected fields that depend on it.2349The endpoint that received the request rejected one of the request's fields as an input it doesn't accept, and the endpoint's message names that field.
2326 2350
2327```text theme={null}2351```text theme={null}
2328API Error: 400 ... Extra inputs are not permitted ... context_management2352API Error: 400 ... context_management: Extra inputs are not permitted
2329```2353```
2330 2354
2331Claude Code sends beta-only fields such as `context_management` alongside an `anthropic-beta` header that enables them. When a gateway forwards the body but drops the header, the API sees fields it doesn't recognize.2355Claude Code sends beta-only fields such as `context_management` and tool search's `defer_loading` alongside an `anthropic-beta` header that enables them. When a gateway forwards the body but drops the header, the API sees fields it doesn't recognize. When a gateway forwards an Anthropic-format body to an upstream with a different schema, that upstream rejects such fields. [Beta header and body field pairs](/docs/en/llm-gateway-protocol#beta-header-and-body-field-pairs) covers both cases.
2332 2356
2333**What to do:**2357**What to do:**
2334 2358
2335* Configure your gateway to forward the `anthropic-beta` header. See [feature pass-through](/docs/en/llm-gateway-protocol#feature-pass-through) for what gateways must forward.2359* If the message contains `context_management: Extra inputs are not permitted`, [update Claude Code](/docs/en/setup#update-claude-code) to v2.1.304 or later. After you update, Claude Code answers this rejection by sending the request once more without the field. Before v2.1.304, Claude Code didn't retry this rejection.
2360* If the message names `defer_loading` or `tool_reference`, update Claude Code or turn [tool search](/docs/en/mcp#tool-search-through-a-gateway) off:
2361 * [Update Claude Code](/docs/en/setup#update-claude-code) to v2.1.305 or later. Claude Code then answers this rejection by sending the request once more without tool search fields.
2362 * If you can't update, or the message still appears after you do, set `ENABLE_TOOL_SEARCH=false` in your shell or your [settings `env`](/docs/en/settings-reference#env). If a settings file already sets the variable, change it there, since that value [can override a shell export](/docs/en/settings-reference#how-env-values-interact-with-your-shell). Claude Code then loads MCP tools upfront.
2363* If your gateway drops the header, configure it to forward the `anthropic-beta` header. See [feature pass-through](/docs/en/llm-gateway-protocol#feature-pass-through) for what gateways must forward.
2336* As a fallback, set [`CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS=1`](/docs/en/env-vars) before launching. [Disable pre-release capabilities](/docs/en/llm-gateway-protocol#disable-pre-release-capabilities) covers the exact scope.2364* As a fallback, set [`CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS=1`](/docs/en/env-vars) before launching. [Disable pre-release capabilities](/docs/en/llm-gateway-protocol#disable-pre-release-capabilities) covers the exact scope.
2337 2365
2338Before v2.1.302, this error could also name the field `diagnostics` with no gateway involved. It appeared when a session that had already used the Anthropic API moved to Amazon Bedrock or Google Cloud's Agent Platform without a restart. For example, a session you left open moves to Bedrock when you run [`/setup-bedrock`](/docs/en/commands) in another Claude Code window. On those versions, run `/clear` in that session or restart it.2366Before v2.1.302, this error could also name the field `diagnostics` with no gateway involved. It appeared when a session that had already used the Anthropic API moved to Amazon Bedrock or Google Cloud's Agent Platform without a restart. For example, a session you left open moves to Bedrock when you run [`/setup-bedrock`](/docs/en/commands) in another Claude Code window. On those versions, run `/clear` in that session or restart it.
2410* Use an alias such as `sonnet` or `opus` instead of a full versioned ID. Aliases resolve to a maintained default so they don't go stale. See [Model configuration](/docs/en/model-config).2438* Use an alias such as `sonnet` or `opus` instead of a full versioned ID. Aliases resolve to a maintained default so they don't go stale. See [Model configuration](/docs/en/model-config).
2411* If the wrong model keeps coming back in the CLI, a stale ID is set somewhere. Check the places you can set a model in [priority order](/docs/en/model-config#setting-your-model) and remove the stale value.2439* If the wrong model keeps coming back in the CLI, a stale ID is set somewhere. Check the places you can set a model in [priority order](/docs/en/model-config#setting-your-model) and remove the stale value.
2412* Claude Code reports an expired claude.ai login as [Login expired](#login-expired), not as this error. Before v2.1.206, an expired login that could no longer be refreshed failed every model with this error; run `/login` if you see that on an older version.2440* Claude Code reports an expired claude.ai login as [Login expired](#login-expired), not as this error. Before v2.1.206, an expired login that could no longer be refreshed failed every model with this error; run `/login` if you see that on an older version.
2441* If [`ANTHROPIC_BASE_URL`](/docs/en/env-vars) points at an [LLM gateway](/docs/en/llm-gateway) and every model fails this way, check whether your gateway is rejecting the tool search fields Claude Code sends by default. To check, set `ENABLE_TOOL_SEARCH=false` as [Tool search through an LLM gateway](/docs/en/mcp#tool-search-through-a-gateway) describes, then try again. Before v2.1.304, tool search was off behind a gateway unless you turned it on with `ENABLE_TOOL_SEARCH`.
2413* For Google Cloud's Agent Platform deployments, see [Google Cloud's Agent Platform troubleshooting](/docs/en/google-vertex-ai#troubleshooting).2442* For Google Cloud's Agent Platform deployments, see [Google Cloud's Agent Platform troubleshooting](/docs/en/google-vertex-ai#troubleshooting).
2414 2443
2415### Model is not a recognized model id2444### Model is not a recognized model id
2616Other messages report the same condition:2645Other messages report the same condition:
2617 2646
2618* In an interactive session, a notification reads `Advisor will not activate on the main model (advisor is less capable); subagents may still use it and may use more tokens · /advisor`.2647* In an interactive session, a notification reads `Advisor will not activate on the main model (advisor is less capable); subagents may still use it and may use more tokens · /advisor`.
2619* At launch with the `--advisor` flag, a warning reads `"<advisor>" cannot advise "<main model>" (the advisor must be at least as capable as the main model). The advisor will not be used for the main model.` and the session starts anyway.2648* When you launch with the `--advisor` flag in [non-interactive mode](/docs/en/headless), and the run isn't a [background session](/docs/en/agent-view), a warning goes to stderr and the run starts anyway. It reads `"<advisor>" cannot advise "<main model>" (the advisor must be at least as capable as the main model). The advisor will not be used for the main model.` An interactive session writes that warning only to the [debug log](#configuration-warnings).
2620 2649
2621**What to do:**2650**What to do:**
2622 2651
4409* Run `claude` in the folder the [debug log](/docs/en/debug-your-config) names and accept the trust dialog. The definition is re-applied the next time Claude Code brings the teammate back; you don't need to restart the lead session4438* Run `claude` in the folder the [debug log](/docs/en/debug-your-config) names and accept the trust dialog. The definition is re-applied the next time Claude Code brings the teammate back; you don't need to restart the lead session
4410* Or set the `hasTrustDialogAccepted` entry to `true` in `~/.claude.json`, using the exact `projects["<path>"]` key the debug log prints4439* Or set the `hasTrustDialogAccepted` entry to `true` in `~/.claude.json`, using the exact `projects["<path>"]` key the debug log prints
4411 4440
4441### Only a teammate can approve a shutdown request
4442
4443Claude Code refused an [agent team](/docs/en/agent-teams) shutdown approval because the session that sent it isn't running as a teammate. Nothing was sent and nothing shut down. The refusal appears in the sending session's tool result:
4444
4445```text wrap theme={null}
4446Only a teammate can approve a shutdown request, and this session is not running as one. Nothing was sent. To shut down a teammate, send that teammate a shutdown_request message. The teammate is the one that approves it.
4447```
4448
4449**What to do:**
4450
4451* If the refusal appeared in your lead session, first check whether the teammate has since shut down. The refusal goes back to Claude with the instruction to send a `shutdown_request` instead, and the turn continues. In an interactive lead session, `<name> has shut down.` appears when the teammate approves a shutdown request.
4452* If the teammate is still running, ask the lead at its prompt to shut the teammate down, for example `Ask the researcher teammate to shut down`. See [Shut down teammates](/docs/en/agent-teams#shut-down-teammates).
4453* If a former teammate you resumed yourself with `claude --resume` got this refusal, nothing was sent. To end that session, run `/exit` at its prompt.
4454
4412### Message too large for cross-session delivery4455### Message too large for cross-session delivery
4413 4456
4414Claude's [cross-session message](/docs/en/cross-session-messaging) to another of your sessions on this machine was too long to send. Claude Code refused it, and the receiving session got nothing. The refusal appears in the sending session's tool result, not as a banner in your terminal. It names both sizes and how to make the message fit:4457Claude's [cross-session message](/docs/en/cross-session-messaging) to another of your sessions on this machine was too long to send. Claude Code refused it, and the receiving session got nothing. The refusal appears in the sending session's tool result, not as a banner in your terminal. It names both sizes and how to make the message fit:
4542 4585
4543### Disk quota or temp filesystem is full4586### Disk quota or temp filesystem is full
4544 4587
4545Claude Code saves each Bash and PowerShell command's output to a file under its temp directory. When a command exits with a nonzero code and no output at all, Claude Code checks whether the filesystem holding that file is out of space or inodes, or whether your disk quota on it is used up. If so, a diagnostic appears in the command's result in place of the empty output:4588Claude Code saves each Bash and PowerShell command's output to a file under its temp directory. These messages in a command's result mean that the filesystem holding that directory, or your disk quota on it, is full.
4589
4590If Claude Code can't create that file because the filesystem or your quota on it is full, your command doesn't start, and the result reads:
4591
4592```text wrap theme={null}
4593The command was not started: this computer's disk is full (ENOSPC), so the file that captures command output could not be created. Nothing ran. If this is the user's own computer, tell them before deleting anything to make room.
4594```
4595
4596When a command exits with a nonzero code and no output at all, Claude Code checks the filesystem holding the output file. If that filesystem is out of space or inodes, or your disk quota on it is used up, a diagnostic replaces the empty output in the command's result:
4546 4597
4547```text wrap theme={null}4598```text wrap theme={null}
4548Your disk quota is full on the filesystem with Claude Code's temp directory /private/tmp/claude-501/-Users-you-my-project/1f0e62dc-4b0a-4f5e-9c2d-8a7b6c5d4e3f/tasks (EDQUOT), so any output this command printed was lost, and it may have failed because it could not write. Delete files you no longer need there, or restart Claude Code with CLAUDE_CODE_TMPDIR set to a directory on another filesystem.4599Your disk quota is full on the filesystem with Claude Code's temp directory /private/tmp/claude-501/-Users-you-my-project/1f0e62dc-4b0a-4f5e-9c2d-8a7b6c5d4e3f/tasks (EDQUOT), so any output this command printed was lost, and it may have failed because it could not write. Delete files you no longer need there, or restart Claude Code with CLAUDE_CODE_TMPDIR set to a directory on another filesystem.
4553* `Your disk quota is full ... (EDQUOT)`: your own quota on that filesystem is used up. A quota can be full while the filesystem still shows free space4604* `Your disk quota is full ... (EDQUOT)`: your own quota on that filesystem is used up. A quota can be full while the filesystem still shows free space
4554* `The filesystem with Claude Code's temp directory ..., or your disk quota on it, is full (ENOSPC)`: the filesystem, or your quota on it, has no space left4605* `The filesystem with Claude Code's temp directory ..., or your disk quota on it, is full (ENOSPC)`: the filesystem, or your quota on it, has no space left
4555* `Command output was lost: the temp filesystem at ... is full` or `... is out of inodes`: the filesystem has almost no free space left, or is running out of inodes4606* `Command output was lost: the temp filesystem at ... is full` or `... is out of inodes`: the filesystem has almost no free space left, or is running out of inodes
4607* `The command was not started: this computer's disk is full (ENOSPC)` or `... (EDQUOT)`: the filesystem has no space left for the output file, or with `EDQUOT` your own quota on it is used up
4556 4608
4557**What to do:**4609**What to do:**
4558 4610
4559* Delete files you no longer need on the filesystem that holds Claude Code's temp directory. For `EDQUOT`, delete files that count against your own quota. For `out of inodes`, delete many files rather than a few large ones, since each file takes one inode whatever its size4611* Delete files you no longer need on the filesystem that holds Claude Code's temp directory. If your message doesn't name that directory, look up its default location in the [`CLAUDE_CODE_TMPDIR`](/docs/en/env-vars) reference. For `EDQUOT`, delete files that count against your own quota. For `out of inodes`, delete many files rather than a few large ones, since each file takes one inode whatever its size
4560* Or restart Claude Code with [`CLAUDE_CODE_TMPDIR`](/docs/en/env-vars) set to a directory on a filesystem with room4612* Or restart Claude Code with [`CLAUDE_CODE_TMPDIR`](/docs/en/env-vars) set to a directory on a filesystem with room
4561* Then have Claude run the command again. The output it printed was lost, not truncated4613* Then have Claude run the command again
4614
4615Before v2.1.305, a command that Claude Code couldn't start because the disk was full returned only the operating system's error text. That text read, for example, `ENOSPC: no space left on device, open <path>.output`.
4562 4616
4563### File is not valid UTF-84617### File is not valid UTF-8
4564 4618
5225 5279
5226## Configuration warnings5280## Configuration warnings
5227 5281
5228Claude Code writes most of these messages to stderr, not into the conversation, and writes most of them at startup. An entry says so when its message appears somewhere else, such as in the debug log or as a startup notice in the conversation view, or at another time, such as the [unrecognized-model diagnostic line](#unrecognized-model-id-on-a-request) at request time.5282To find a warning that Claude Code gave at startup in an interactive session, look below the startup banner first, then in the debug log:
5283
5284* **Row below the startup banner**: some warnings about [permission rules](/docs/en/permissions), such as [`⚠ Deny rule Bsh matches no tool · check for typos`](#matches-no-tool)
5285* **Debug log**: Claude Code keeps this log only in debug mode. To read the warnings in it, start Claude Code from your shell with `claude --debug-file <path>`, then open the file at that path.
5286
5287In [non-interactive mode](/docs/en/headless) with text output, outside a [background session](/docs/en/agent-view), look for startup warnings on stderr.
5288
5289Before v2.1.305, look on stderr in an interactive session as well, including for the permission-rule warnings that now show as rows below the banner.
5229 5290
5230### Claude Code exited after an unrecoverable interface error5291### Claude Code exited after an unrecoverable interface error
5231 5292
5345 5406
5346### Working directory is a network path5407### Working directory is a network path
5347 5408
5348Claude Code doesn't add network paths as working directories. Looking up a network path can contact the host it names, and on Windows that contact can send the host your credentials, so Claude Code refuses the path without looking it up. You see this message when you run `/add-dir` with such a path, or as a warning at startup. When it appears at startup, Claude Code starts without that directory.5409Claude Code doesn't add network paths as working directories. Looking up a network path can contact the host it names, and on Windows that contact can send the host your credentials. Claude Code therefore refuses the path without looking it up.
5410
5411You see this message when you run `/add-dir` with a network path:
5349 5412
5350```text theme={null}5413```text theme={null}
5351\\server\share is a network path, which cannot be added as a working directory. On Windows, map the share to a drive letter and pass it at launch with --add-dir (a drive letter added mid-session does not yet carry remote-read trust).5414\\server\share is a network path, which cannot be added as a working directory. On Windows, map the share to a drive letter and pass it at launch with --add-dir (a drive letter added mid-session does not yet carry remote-read trust).
5352```5415```
5353 5416
5417When you pass one at startup instead, with `--add-dir` or in `permissions.additionalDirectories`, Claude Code starts without that directory. In an interactive session you don't see the message at startup, because it goes to the debug log with the other [startup warnings](#configuration-warnings).
5418
5354Paths that Claude Code refuses this way include:5419Paths that Claude Code refuses this way include:
5355 5420
5356* UNC shares such as `\\server\share`5421* UNC shares such as `\\server\share`
5596 5661
5597Before v2.1.260, Claude Code reported a rule with unmatched parentheses as `Mismatched parentheses`.5662Before v2.1.260, Claude Code reported a rule with unmatched parentheses as `Mismatched parentheses`.
5598 5663
5599### Is not matched by file permission checks5664### Matches no tool · check for typos
5600 5665
5601Claude Code found a `Write`, `NotebookEdit`, `MultiEdit`, or `Glob` [permission rule](/docs/en/permissions#read-and-edit) with a path in one of your [settings files](/docs/en/settings#where-settings-live), in [managed settings](/docs/en/managed-settings), or in a `--allowedTools`, `--disallowedTools`, or `--settings` flag value. It checks file permissions against `Edit` and `Read` rules only, so it never consults a path rule that names one of the other file tools. It keeps the rule and changes nothing else; the warning names the rule, its source in parentheses, and the replacement to write:5666This row below the startup banner names a deny or ask [permission rule](/docs/en/permissions) whose tool name Claude Code doesn't recognize, so the rule doesn't apply to the tool you meant. A deny rule written as `Bsh` isn't applied to Bash, and neither is one that uses the label a tool shows in the transcript instead of its canonical name. The row repeats the rule as you wrote it:
5602 5667
5603```text theme={null}5668```text theme={null}
5604Permission deny rule (.claude/settings.json): Write(docs/**) is not matched by file permission checks — only Edit(path) rules are. Use Edit(docs/**) instead (Edit rules cover all file-editing tools).5669⚠ Deny rule Bsh matches no tool · check for typos
5670```
5671
5672In [non-interactive mode](/docs/en/headless) with text output, Claude Code prints the warning on stderr in its longer form instead:
5673
5674```text theme={null}
5675Permission deny rule "Bsh" matches no known tool — check for typos.
5605```5676```
5606 5677
5607**What to do:**5678**What to do:**
5608 5679
5680* Correct the tool name in the rule, for example `Bash` in place of `Bsh`. Look up canonical names in the [tools reference](/docs/en/tools-reference).
5681* To find where the rule is defined, run `/permissions` in the session and select the rule. Its details say where it comes from, such as `From project local settings`, or `From CLI argument` for a `--disallowedTools` value.
5682* If the details read `From enterprise managed settings`, forward the row to whoever maintains your [managed settings](/docs/en/managed-settings), since you can't change that rule yourself.
5683
5684In a [background session](/docs/en/agent-view) or with `--output-format json` or `stream-json`, Claude Code writes the warning to the debug log instead of stderr. Run with `--debug` to capture it at `~/.claude/debug/<session-id>.txt`.
5685
5686Before v2.1.305, an interactive session printed the longer form on stderr at startup instead of showing the row.
5687
5688### Is not matched by file permission checks
5689
5690Claude Code found a `Write`, `NotebookEdit`, `MultiEdit`, or `Glob` [permission rule](/docs/en/permissions#read-and-edit) with a path in one of your [settings files](/docs/en/settings#where-settings-live), in [managed settings](/docs/en/managed-settings), or in a `--allowedTools`, `--disallowedTools`, or `--settings` flag value. It checks file permissions against `Edit` and `Read` rules only, so it never consults a path rule that names one of the other file tools. It keeps the rule and changes nothing else.
5691
5692* **Interactive session**: the full warning goes to the [debug log](#configuration-warnings), and a deny or ask rule of this shape also shows as this row below the startup banner:
5693
5694 ```text theme={null}
5695 ⚠ Deny rule Write(docs/**) never matches · use Edit(docs/**)
5696 ```
5697
5698 An allow rule of this shape gets no row. To find where the rule is defined, run `/permissions` in the session and select the rule.
5699* **[Non-interactive mode](/docs/en/headless) with text output**: the full warning on stderr names the rule, its source in parentheses, and the replacement to write:
5700
5701 ```text theme={null}
5702 Permission deny rule (.claude/settings.json): Write(docs/**) is not matched by file permission checks — only Edit(path) rules are. Use Edit(docs/**) instead (Edit rules cover all file-editing tools).
5703 ```
5704* **[Background session](/docs/en/agent-view), or `--output-format json` or `stream-json`**: the full warning goes to the debug log instead of stderr, so machine-read output stays clean. Run with `--debug` to capture it at `~/.claude/debug/<session-id>.txt`.
5705
5706**What to do:**
5707
5609* Replace `Write(path)`, `NotebookEdit(path)`, and legacy `MultiEdit(path)` rules with `Edit(path)`. `Edit` rules cover all file-editing tools.5708* Replace `Write(path)`, `NotebookEdit(path)`, and legacy `MultiEdit(path)` rules with `Edit(path)`. `Edit` rules cover all file-editing tools.
5610* Except in `--allowedTools`, where Claude Code accepts a `Glob` rule without warning, replace `Glob(path)` rules with `Read(path)`.5709* Except in `--allowedTools`, where Claude Code accepts a `Glob` rule without warning, replace `Glob(path)` rules with `Read(path)`.
5611* Fix the rule at the source the warning names in parentheses: a settings file path, or the flag itself for `--allowed-tools` and `--disallowed-tools`. A `claude-settings-<hash>.json` path that doesn't exist on disk stands for an inline `--settings` value. Fix the JSON you pass to that flag.5710* Fix the rule at its source, which the full warning names in parentheses: a settings file path, or the flag itself for `--allowed-tools` and `--disallowed-tools`. A `claude-settings-<hash>.json` path that doesn't exist on disk stands for an inline `--settings` value. Fix the JSON you pass to that flag.
5612* Leave bare tool-name rules such as `Write` or `Glob` alone. Claude Code matches them at the [tool level](/docs/en/permissions#match-all-uses-of-a-tool) and doesn't warn about them.5711* Leave bare tool-name rules such as `Write` or `Glob` alone. Claude Code matches them at the [tool level](/docs/en/permissions#match-all-uses-of-a-tool) and doesn't warn about them.
5613* If the source reads `managed policy settings`, forward the warning to whoever maintains your managed settings, since you can't clear it yourself.5712* For a rule from managed settings, the full warning's parentheses read `managed policy settings` and the rule's details in `/permissions` read `From enterprise managed settings`. Forward that warning to whoever maintains your managed settings, since you can't clear it yourself.
5614 5713
5615In a [background session](/docs/en/agent-view) or with `--output-format json` or `stream-json`, Claude Code writes the warning to the debug log instead of stderr, so machine-read output stays clean. Run with `--debug` to capture it at `~/.claude/debug/<session-id>.txt`. Before v2.1.210, Claude Code accepted these rules without a warning.5714Before v2.1.210, Claude Code accepted these rules without a warning. Before v2.1.305, an interactive session printed the full warning on stderr.
5616 5715
5617### Has a wildcard before the rest of the command5716### Has a wildcard before the rest of the command
5618 5717
5619Claude Code found a `Bash` allow rule whose `*` comes before a later word that determines which command it is, such as `Bash(git * main)` or `Bash(git -C * status *)`, in one of your [settings files](/docs/en/settings#where-settings-live), in [managed settings](/docs/en/managed-settings), or in an `--allowedTools` or `--settings` flag value. The `*` matches any text, including options inserted at that position: `Bash(git * main)` also approves `git -c core.fsmonitor=<script> diff main`, where `-c` makes git run a program the command names. [Wildcard patterns](/docs/en/permissions#wildcard-patterns) shows the matching rules.5718Claude Code found a `Bash` allow rule whose `*` comes before a later word that determines which command it is, such as `Bash(git * main)` or `Bash(git -C * status *)`, in one of your [settings files](/docs/en/settings#where-settings-live), in [managed settings](/docs/en/managed-settings), or in an `--allowedTools` or `--settings` flag value. The `*` matches any text, including options inserted at that position: `Bash(git * main)` also approves `git -c core.fsmonitor=<script> diff main`, where `-c` makes git run a program the command names. [Wildcard patterns](/docs/en/permissions#wildcard-patterns) shows the matching rules.
5620 5719
5621The warning exists so you can narrow a rule whose wildcard is broader than you intended. Claude Code keeps the rule and changes nothing about how it matches; the warning names the rule and its source in parentheses:5720The warning exists so you can narrow a rule whose wildcard is broader than you intended. Claude Code keeps the rule and changes nothing about how it matches. In [non-interactive mode](/docs/en/headless) with text output, the warning on stderr names the rule and its source in parentheses:
5622 5721
5623```text theme={null}5722```text theme={null}
5624Permission allow rule (.claude/settings.json): Bash(git -C * status *) has a wildcard before the rest of the command, so it also matches any options inserted at that position and approves them without a prompt. For git, options such as -c and --exec-path can run arbitrary commands. Replace that * with the exact value you mean, or only use * after the subcommand (for example Bash(git status *)).5723Permission allow rule (.claude/settings.json): Bash(git -C * status *) has a wildcard before the rest of the command, so it also matches any options inserted at that position and approves them without a prompt. For git, options such as -c and --exec-path can run arbitrary commands. Replace that * with the exact value you mean, or only use * after the subcommand (for example Bash(git status *)).
5625```5724```
5626 5725
5726In an interactive session, the full warning goes to the [debug log](#configuration-warnings) and the rule shows as this row below the startup banner:
5727
5728```text theme={null}
5729⚠ Allow rule Bash(git -C * status *) allows any option at its * · use an exact value
5730```
5731
5732To find where the rule is defined, run `/permissions` in the session and select the rule.
5733
5627**What to do:**5734**What to do:**
5628 5735
5629* Replace the `*` before the subcommand with the exact value you mean: `Bash(git checkout main)` in place of `Bash(git * main)`.5736* Replace the `*` before the subcommand with the exact value you mean: `Bash(git checkout main)` in place of `Bash(git * main)`.
5630* Move every `*` after the subcommand: `Bash(git status *)` in place of `Bash(git -C * status *)`. Write one rule per subcommand you want to allow.5737* Move every `*` after the subcommand: `Bash(git status *)` in place of `Bash(git -C * status *)`. Write one rule per subcommand you want to allow.
5631* Fix the rule at the source the warning names in parentheses: a settings file path, or the `--allowed-tools` flag itself. A `claude-settings-<hash>.json` path that doesn't exist on disk stands for an inline `--settings` value. Fix the JSON you pass to that flag.5738* Fix the rule at its source, which the full warning names in parentheses: a settings file path, or the `--allowed-tools` flag itself. A `claude-settings-<hash>.json` path that doesn't exist on disk stands for an inline `--settings` value. Fix the JSON you pass to that flag.
5632* If the source reads `managed policy settings`, forward the warning to whoever maintains your managed settings, since you can't clear it yourself.5739* For a rule from managed settings, the full warning's parentheses read `managed policy settings` and the rule's details in `/permissions` read `From enterprise managed settings`. Forward that warning to whoever maintains your managed settings, since you can't clear it yourself.
5633 5740
5634In a [background session](/docs/en/agent-view) or with `--output-format json` or `stream-json`, Claude Code writes the warning to the debug log instead of stderr, so machine-read output stays clean. Run with `--debug` to capture it at `~/.claude/debug/<session-id>.txt`. Before v2.1.246, Claude Code accepted these rules without a warning.5741In a [background session](/docs/en/agent-view) or with `--output-format json` or `stream-json`, Claude Code writes the warning to the debug log instead of stderr, so machine-read output stays clean. Run with `--debug` to capture it at `~/.claude/debug/<session-id>.txt`. Before v2.1.246, Claude Code accepted these rules without a warning.
5635 5742
5743Before v2.1.305, an interactive session printed the full warning on stderr instead of showing a row.
5744
5636### Denying Bash also turns off the PowerShell tool5745### Denying Bash also turns off the PowerShell tool
5637 5746
5638You removed the whole Bash tool, for example with `--disallowedTools Bash` or with a bare `Bash` or `Bash(*)` [deny rule](/docs/en/permissions#match-all-uses-of-a-tool) in one of your settings files. On Windows with Git Bash installed, [denying Bash also turns the PowerShell tool off](/docs/en/tools-reference#bash-deny-rules-also-turn-off-the-powershell-tool), so the session starts with no shell tool. Claude Code prints this warning at startup:5747You removed the whole Bash tool, for example with `--disallowedTools Bash` or with a bare `Bash` or `Bash(*)` [deny rule](/docs/en/permissions#match-all-uses-of-a-tool) in one of your settings files. On Windows with Git Bash installed, [denying Bash also turns the PowerShell tool off](/docs/en/tools-reference#bash-deny-rules-also-turn-off-the-powershell-tool), so the session starts with no shell tool.
5748
5749In [non-interactive mode](/docs/en/headless) with text output, Claude Code prints this warning on stderr at startup, and an interactive session writes it only to the [debug log](#configuration-warnings):
5639 5750
5640```text theme={null}5751```text theme={null}
5641Denying Bash also turns off the PowerShell tool, so Claude has neither. To use PowerShell, set CLAUDE_CODE_USE_POWERSHELL_TOOL=1.5752Denying Bash also turns off the PowerShell tool, so Claude has neither. To use PowerShell, set CLAUDE_CODE_USE_POWERSHELL_TOOL=1.
5648 5759
5649In a [background session](/docs/en/agent-view) or with `--output-format json` or `stream-json`, Claude Code writes the warning to the debug log instead of stderr. Run with `--debug` to capture it at `~/.claude/debug/<session-id>.txt`. Before v2.1.287, Claude Code turned the PowerShell tool off the same way without printing a warning.5760In a [background session](/docs/en/agent-view) or with `--output-format json` or `stream-json`, Claude Code writes the warning to the debug log instead of stderr. Run with `--debug` to capture it at `~/.claude/debug/<session-id>.txt`. Before v2.1.287, Claude Code turned the PowerShell tool off the same way without printing a warning.
5650 5761
5762Before v2.1.305, an interactive session printed the warning on stderr at startup too.
5763
5651### crossSessionInbound must be one of accept, hold, refuse5764### crossSessionInbound must be one of accept, hold, refuse
5652 5765
5653A settings file sets [`crossSessionInbound`](/docs/en/settings-reference#crosssessioninbound) to a value Claude Code doesn't recognize, such as the typo `"reject"`. The warning's second sentence depends on which file holds the value; in a user, project, local, or `--settings` file it reads:5766A settings file sets [`crossSessionInbound`](/docs/en/settings-reference#crosssessioninbound) to a value Claude Code doesn't recognize, such as the typo `"reject"`. The warning's second sentence depends on which file holds the value; in a user, project, local, or `--settings` file it reads:
5697* If the message names a model ID this version doesn't recognize, run `claude update`. A version that recognizes the ID as a 1M-context model enforces the limit without further configuration.5810* If the message names a model ID this version doesn't recognize, run `claude update`. A version that recognizes the ID as a 1M-context model enforces the limit without further configuration.
5698* If you want the session to use the model's full window instead, unset `CLAUDE_CODE_DISABLE_1M_CONTEXT`; the warning reports only that the 200K limit isn't enforced5811* If you want the session to use the model's full window instead, unset `CLAUDE_CODE_DISABLE_1M_CONTEXT`; the warning reports only that the 200K limit isn't enforced
5699 5812
5700In a [background session](/docs/en/agent-view) or with `--output-format json` or `stream-json`, Claude Code writes the warning to the debug log instead of stderr.5813In [non-interactive mode](/docs/en/headless) with text output, this warning appears on stderr. In an interactive session, a [background session](/docs/en/agent-view), or with `--output-format json` or `stream-json`, it goes to the debug log instead. Run with `--debug` to capture it at `~/.claude/debug/<session-id>.txt`.
5814
5815Before v2.1.305, an interactive session printed the warning on stderr at startup.
5701 5816
5702### Unrecognized model ID on a request5817### Unrecognized model ID on a request
5703 5818
