claude-code-system-prompts - All parts of Claude Code's system prompt, 27 built

项目说明

<di="left">

Check out Piebald

We' Piebald, the ultimate agentic AI de experience. \
Download it and try it out for free!

<sub>Scroll down for Claude 代码's system prompts. :point_down:</sub>

</div>
</div>

<di="left">
<a href="
<picture>
<源码 media="(prefers-color-scheme: dark)" srcset="
<源码 media="(prefers-color-scheme: light)" srcset="
<img alt="hero" width="800" src="
</picture>
</a>
</div>

Claude Code System Prompts

![Mentioned in Awesome Claude Code]()

> [!tip]
> NEW (June 12, 2026): We' expanded this list with many more of Claude 代码's prompts—from 350 to 515 (+165)—our most complete co yet.

This repository contains an up-to-date list of all Claude 代码's system prompts and their associated token counts as of Claude 代码 v2.1.289 (October 3rd, 2026). It also contains a CHANGELOG.md for the system prompts across 305 since v2.0.14. From the team behind <img src=" width="15"> Piebald.

This repository is updated within minutes of each Claude 代码 release. See the changelog, and follow on X for a summary of the system prompt changes in each release.

> [!note]
> ⭐ Star this repository to get notified about new Claude 代码 . For each new Claude 代码 , we create a release on GitHub, which will notify all users who' the repository.

---

Why multiple "system prompts?"

Claude 代码 doesn't just have one single string for its system prompt.

Instead, there are:

Large portions conditionally added depending on the en and configs.

Descriptions for builtin tools like Write, Bash, and TodoWrite, and some are fairly large.

Separate system prompts for builtin agents like Explore and Plan.

Numerous AI-powered utility functions, such as con compaction, CLAUDE.md generation, session title generation, etc. featuring their own systems prompts.

The result—500+ strings that are constantly changing and moving within a very large minified JS file.

> [!important]
> Want to modify a particular piece of the system prompt in your own Claude Code installation? Use tweakcc. It—
> - lets you customize the indi pieces of the system prompt as markdown files, and then
> - patches your npm-based or nati) Claude 代码 installation with them, and also
> - provides diffing and conflict management for when both you and Anthropic ha modifications to the same prompt file.

Extraction

This repository contains the system prompts extracted using a script from the latest npm of Claude 代码. As they're extracted directly from Claude 代码's compiled 源码代码, they're guaranteed to be exactly what Claude 代码 uses. If you use tweakcc to customize the system prompts, it works in a similar way—it patches the exact same strings in your local installation as are extracted into this repository.

Prompts

Note that some prompts contain interpolated bits such as builtin 工具 name references, lists of a sub agents, and other context-specific , so the actual counts in a particular Claude 代码 session will differ slightly—likely not beyond ±20 tokens, however.

Agent Prompts

Sub-agents and utilities.

#### Sub-agents

Agent Prompt: Explore (862 tks) - System prompt for the Explore subagent.

Agent Prompt: Plan mode (enhanced) (1066 tks) - Enhanced prompt for the Plan subagent.

#### Creation Assistants

Agent Prompt: CLAUDE.md creation (631 tks) - Instructs the /init command to analyze a codebase and create or improve a CLAUDE.md file.

Agent Prompt: Status line setup (4815 tks) - System prompt for the statusline-setup agent that configures status line display.

#### Slash Commands

Agent Prompt: /batch slash command (1524 tks) - Instructions for orchestrating a large, parallelizable change across a codebase.

Agent Prompt: /代码-review GitLab comment posting (312 tks) - Directs 代码 review findings to a general GitLab merge request note when the comment flag is enabled.

Agent Prompt: /代码-review inline gap sweep phase (105 tks) - Adds a final same-context sweep for defects missed by an inline 代码 review when subagents are una.

Agent Prompt: /代码-review minimal mode (632 tks) - Minimal /代码-review prompt that performs one careful diff pass and reports up to fifteen concrete correctness findings.

Agent Prompt: /代码-review part 1 base finder angles (145 tks) - Line-by-line diff scan instructions for the /代码-review slash command's finder-angle phase.

Agent Prompt: /代码-review part 10 ReportFindings output format (448 tks) - Output-format instructions for /代码-review runs that report findings once through the ReportFindings 工具 with capped, se findings.

Agent Prompt: /代码-review part 2 low effort minimum findings mode (740 tks) - Low-effort /代码-review prompt that reads the diff once, targets at least min(files_changed, 4) hunk- runtime correctness findings, and performs one extra pass when short.

Agent Prompt: /代码-review part 2 low effort mode (778 tks) - Low-effort /代码-review prompt that reads the diff once and returns up to four hunk- runtime correctness findings.

Agent Prompt: /代码-review part 3 extra-high and maximum effort modes (553 tks) - Extra-high and maximum-effort /代码-review prompt that runs fi angles, one-vote , a gap sweep, and capped JSON findings.

Agent Prompt: /代码-review part 4 three-state phase (98 tks) - phase for /代码-review that asks one agent to classify each candidate as confirmed, plausible, or refuted.

Agent Prompt: /代码-review part 5 recall-biased phase (175 tks) - Recall-biased /代码-review phase that treats realistic uncertain findings as plausible unless 代码 refutes them.

Agent Prompt: /代码-review part 6 medium effort mode (466 tks) - Medium-effort /代码-review prompt that favors precision with three finder angles, one-vote , and up to eight JSON findings.

Agent Prompt: /代码-review part 7 high effort mode (511 tks) - High-effort /代码-review prompt that favors recall with three finder angles, recall-biased , and up to ten JSON findings.

Agent Prompt: /代码-review part 8 GitHub comment posting (152 tks) - Optional /代码-review instructions for posting findings as GitHub inline PR comments when --comment is passed.

Agent Prompt: /代码-review part 9 fix application (250 tks) - Optional /代码-review instructions for applying findings to the working tree when --fix is passed.

Agent Prompt: /代码-review una inline mode (403 tks) - Builds a single-pass inline 代码-review prompt that gathers a diff, e configured angles, deduplicates findings, optionally sweeps gaps, and reports capped results when the Agent 工具 is una.

Agent Prompt: /rename auto-generate session name (80 tks) - Prompt used by /rename (no args) to auto-generate a kebab-case session name from con context.

Agent Prompt: /schedule slash command (4469 tks) - Guides the user through scheduling, updating, listing, or running remote Claude 代码 agents on cron triggers via the Anthropic cloud API.

Agent Prompt: /security-review slash command (2521 tks) - Comprehensi review prompt for analyzing 代码 changes with focus on exploitable .

Agent Prompt: /simplify slash command (362 tks) - Instructions for the /simplify slash command that reviews changed 代码 for reuse, simplification, efficiency, and altitude cleanups, then applies the fixes.

Agent Prompt: /simplify una inline mode (621 tks) - Runs the /simplify cleanup workflow inline across reuse, simplification, efficiency, and altitude angles when the Agent 工具 is una.

Agent Prompt: Artifact type creation slash command (quickstart flow) (277 tks) - Directs an Artifact creation slash command to quickstart the requested published type, create the new Artifact, and follow the returned type instructions.

Agent Prompt: Artifact type creation slash command (347 tks) - Directs an Artifact creation slash command to discover the named published Artifact type, create from its type URL, and follow the returned type instructions.

#### Utilities

Agent Prompt: Agent Hook (282 tks) - E agent hook conditions against the codebase and, when a, the con transcript, then returns a structured pass/fail result.

Agent Prompt: Artifact comment completion reply and resolution (256 tks) - Directs an Artifact comment worker to post one completion reply when needed and resolve an open thread after acting unless the con remains active.

Agent Prompt: Artifact comment thread analyst (750 tks) - Instructs a read-only agent to analyze one Artifact comment thread and return a constrained analysis brief for a separate composer.

Agent Prompt: Artifact comment thread background directive (263 tks) - Directs a forked background agent to handle one Artifact comment thread on its own, leaving the main session's pending work alone and treating thread and artifact content as material rather than instructions.

Agent Prompt: Artifact editor thread follow-up (328 tks) - Delivers a tagged Artifact thread message to the acti worker, requiring page-scoped edits and republishing for relevant requests and no changes for unrelated ones.

Agent Prompt: Auto mode rule re (292 tks) - Reviews and critiques user-defined auto mode classifier rules for clarity, completeness, conflicts, and actionability.

Agent Prompt: Away summary generation (73 tks) - Prompts a no-tools away-summary generation run to recap the goal, current task, and next action when the user returns.

Agent Prompt: Background agent state classifier (6237 tks) - Classifies the tail of a background agent transcript as working, blocked, done, or failed and returns concise state JSON.

Agent Prompt: Background job agent instructions (581 tks) - Instructs the built-in background job agent to narrate progress, restate 工具 results, and emit explicit result, needs input, or failed status signals.

Agent Prompt: Claude Code guide (408 tks) - Describes when to use the Claude 代码 guide agent for Claude 代码, Agent SDK, Claude API, and Claude Tag questions.

Agent Prompt: Claude Test author (346 tks) - Subagent definition for the Claude Test plugin's background spec author, which reads the app's 源码 and writes spec drafts into a run folder's proposed/ directory for the person to approve.

Agent Prompt: Claude Test explorer (777 tks) - Subagent definition for the Claude Test plugin's read-only explorer, which maps a web application's 源码 for a browser test suite using only Read, Grep and Glob and must avoid credential and out-of-project paths.

Agent Prompt: Claude guide agent (2975 tks) - Instructs the Claude guide agent to answer documentation-based questions about Claude 代码, the Claude Agent SDK, the Claude API, Claude Tag, plugin e, and skill diagnostics.

Agent Prompt: Coding session title generator (898 tks) - Generates a short noun-phrase title for a coding session from untrusted session content.

Agent Prompt: Con summarization with additional instructions (1380 tks) - Extended summarization prompt with support for custom additional instructions.

Agent Prompt: Con summarization (343 tks) - System prompt for creating detailed con summaries.

Agent Prompt: Coordinator worker instructions (830 tks) - Instructions for worker agents executing coordinator-assigned tasks, co scope control, concurrent branch changes, resumption, failure handling, and coordinator-facing output.

Agent Prompt: Determine which memory files to attach (388 tks) - Agent for determining which memory files to attach for the main agent.

Agent Prompt: Dream memory consolidation (1361 tks) - Instructs an agent to perform a multi-phase memory consolidation pass — orienting on existing memories, gathering recent signal from logs and transcripts, merging updates into topic files, and pruning the index.

Agent Prompt: General purpose agent (63 tks) - Defines a general-purpose agent for researching complex questions, searching 代码, and executing multi-step tasks.

Agent Prompt: General purpose (446 tks) - System prompt for the general-purpose subagent that searches, analyzes, and edits 代码 across a codebase while reporting findings concisely to the caller.

Agent Prompt: General task agent (99 tks) - Instructs a Claude 代码 task agent to complete the user's request fully and report the essential outcome.

Agent Prompt: Hook condition e (stop) (319 tks) - System prompt for e hook conditions, specifically stop conditions, in Claude 代码.

Agent Prompt: Hook condition e (88 tks) - Instructs an agent to judge whether a user-provided hook condition is met.

Agent Prompt: Inherited context for worktree sub-agent (121 tks) - Briefs a sub-agent that it has inherited a parent session's context and is now working in its own isolated git worktree.

Agent Prompt: Onboarding guide draft share link workflow (423 tks) - Adds instructions for sharing the draft ONBOARDING.md before review, then updating the same ShareOnboardingGuide link after the user answers the review questions.

Agent Prompt: Onboarding guide generator (1135 tks) - Co-authors a team onboarding guide (ONBOARDING.md) for new Claude 代码 users by analyzing the creator's usage data, classifying session types, and iterating on the draft collaboratively.

Agent Prompt: PR follow-up cron (PR Steward label check) (278 tks) - PR follow-up cron clause telling the agent not to fix or push a PR carrying PR Steward labels, to tell the user once, and to offer leave-it, one-change, or take-o.

Agent Prompt: PR follow-up cron (332 tks) - Cron prompt for checking a pull request created in the session and fixing failures, comments, or conflicts.

Agent Prompt: Plugin e trust requirement (205 tks) - Requires explicit user trust for a plugin directory before piloting its e with --trust-plugin and otherwise limits work to writing case files.

Agent Prompt: Project thread status card classifier (4299 tks) - Classifies a Project thread as needing a reply, needing approval, done, failed, or working and emits concise owner-facing status-card JSON.

Agent Prompt: Prompt Suggestion Generator v2 (344 tks) - V2 instructions for generating prompt suggestions for Claude 代码.

Agent Prompt: Pull request creation (1207 tks) - Prompt for creating a single GitHub pull request from existing commits with branch, template, attribution, shell-formatting, and git-safety guidance.

Agent Prompt: Quick PR creation (1741 tks) - Streamlined prompt for creating a commit and pull request with pre-populated context.

Agent Prompt: Quick git commit (1368 tks) - Streamlined prompt for creating a single git commit with pre-populated context.

Agent Prompt: Read-only search agent (93 tks) - Defines a read-only search agent for broad fan-out 代码 searches that returns conclusions instead of file dumps.

Agent Prompt: Recent Message Summarization (868 tks) - Agent prompt used for summarizing recent messages.

Agent Prompt: Remote machine auto mode rules (228 tks) - Adds host-declared deny, allow, and en rules to the security monitor prompt for commands that will run on another machine.

Agent Prompt: Schedule action selection (114 tks) - Instructs the cloud scheduling agent to ask the user which schedule action to perform first.

Agent Prompt: Schedule local MCP ser (1145 tks) - Explains that local Claude 代码 MCP servers cannot be attached to cloud routines and tailors Claude.ai connector guidance to why the connector list was una.

Agent Prompt: Security monitor Claude Tag connector writes (542 tks) - Allows the autonomous-action security monitor to accept delegated writes through exact Claude Tag connector 工具 prefixes while preserving all other safety boundaries.

Agent Prompt: Security monitor attached machine call results (269 tks) - Explains how the auto mode security monitor reads _host call lines and result_from lines for calls served by a user-attached machine, and judges sending that machine's output to external destinations as cross-machine data exfiltration.

Agent Prompt: Security monitor for autonomous agent actions (first part) (13991 tks) - Instructs Claude to act as a security monitor that e autonomous coding agent actions against block/allow rules to prevent prompt injection, scope creep, and accidental damage.

Agent Prompt: Security monitor forwarded user turns (761 tks) - Defines how the auto mode classifier authenticates harness-forwarded user turns and weighs direct, relayed, and unattributed messages as action-specific intent, consent, or restrictions.

Agent Prompt: Security monitor host-context line guidance (614 tks) - Explains how the autonomous-action security monitor should interpret live and restored host-context lines when e user intent and consent.

Agent Prompt: Session title and branch generation (626 tks) - Agent for generating succinct session titles and git branch names.

Agent Prompt: Session transcript chunk summary (89 tks) - Instructs an agent to summarize a chunk of a Claude 代码 session transcript concisely.

Agent Prompt: Summarization no-tools guard (107 tks) - Shared prefix for compaction summarization agents that forbids 工具 use and requires plain text analysis and summary blocks.

Agent Prompt: Web fetch agent usage guidance (504 tks) - Explains when and how to delegate URL reading to the built-in web-fetch agent, including binary-file trust boundaries and follow-up messaging.

Agent Prompt: Web reading specialist (905 tks) - System prompt for the built-in web-fetch agent that reads untrusted URL content with WebFetch and returns a focused, 源码-grounded report to its caller.

Agent Prompt: Worker fork (412 tks) - Directi into a forked child agent, telling it to treat the inherited transcript as reference, execute one directi, and report once.

Agent Prompt: Workflow script plain text return note (76 tks) - Appended note telling a workflow script agent that its final text response is parsed as the script return value.

Agent Prompt: Workflow script structured return note (120 tks) - Appended note telling a workflow script agent to return its final answer by calling the structured output 工具 exactly once.

Agent Prompt: Workflow subagent plain text output (154 tks) - Instructs an internal workflow subagent to return its final text as the calling workflow script's parsed result.

Agent Prompt: Workflow subagent structured output (190 tks) - Instructs an internal workflow subagent to return its final answer by calling the StructuredOutput 工具 exactly once with schema-valid input.

Data

The content of template files embedded in Claude 代码.

Data: /auto-mode-setup usage (406 tks) - Defines /auto-mode-setup proposal and re command syntax, request ID ordering, sa , and integrity requirements.

Data: Artifact MCP connector guidance (351 tks) - Explains how Artifact MCP manifests declare claude.ai, built-in, and host connectors and how to disco 服务端 and upstream 工具 names.

Data: Artifact capability pass (271 tks) - Requires one functional pass over a page's session-declared runtime capabilities before handing over its link, assembling the preview, 数据库 read and endpoint checks to run and what to report to the user.

Data: Artifact connector call obser requirement (305 tks) - Explains how to obtain connector argument names from the loaded 工具 schema, learn result shapes from safe real calls, and disclose shapes that cannot be before publishing.

Data: Artifact connector ser guidance (241 tks) - Explains how connector 工具 prefixes map to manifest ser and why in-page calls must use the resol display name.

Data: Artifact decision component HTML skeleton (1117 tks) - Provides the canonical HTML and JSON-island skeleton for embedding one reusable Artifact decision block.

Data: Artifact host MCP ser (420 tks) - Explains how locally configured MCP servers may be declared as Artifact host servers and distinguishes them from Claude.ai connectors and built-in servers.

Data: Artifact publish merged-o file changes notice (481 tks) - 工具-result notice appended when an artifact publish was merged over a newer li, listing files changed, added, or remo since the base , that stale copies must be read again, and how many further files are not listed.

Data: Artifact runtime capability declarations (496 tks) - Defines Artifact runtime capability declaration, carry-forward, clearing, replacement, and contract pinning semantics.

Data: Auto-compact inputs changed event schema (221 tks) - Schema description for worker-resol state events used by thin clients to display the effecti countdown.

Data: Background tasks changed event schema (342 tks) - Schema description for the background_tasks_changed system event and its replace-set semantics.

Data: Built-in gh stand-in api command help (1055 tks) - Usage text for Claude 代码's built-in gh stand-in, which supports only gh api REST requests to GitHub hosts through the session's GitHub proxy, listing its flags and how {owner}, {repo} and {branch} placeholders are filled in.

Data: Chrome browser hints control request (207 tks) - Documents the set_chrome_browser_hints control request that tells a session which connected Claude in Chrome browser to prefer, and that the hints are session-only and ne the li roster.

Data: Claude API reference — PHP (2877 tks) - PHP SDK reference.

Data: Claude API reference — Ruby (2120 tks) - Ruby SDK reference including installation, 客户端 initialization, basic requests, streaming, and beta 工具 runner.

Data: Claude 代码 agent proxy troubleshooting guide (2214 tks) - Troubleshooting guide for Claude 代码's policy-enforcing HTTPS agent proxy, co TLS trust setup, status checks, git, docker, and unsupported traffic.

Data: Claude 代码 gateway customer-routed inference 协议 (2945 tks) - Conditional extension to the Claude 代码 gateway 协议 defining customer-routed inference authentication, forwarding, response hygiene, error recovery, policy blocking, discovery, and endpoint requirements.

Data: Claude Code gateway protocol (7396 tks) - Markdown reference documenting the Claude 代码 gateway wire contract, including OAuth 2.0 device flow, RFC 8414 discovery, Messages API inference, managed settings, model discovery, OTLP telemetry, error en, TLS certificate pinning, and proxying to Bedrock, , and Foundry.

Data: Claude Code li sources (2477 tks) - WebFetch URLs for fetching current Claude 代码 documentation from official sources.

Data: Claude 代码 recent changes reference (1854 tks) - Reference mapping of recently removed or renamed Claude 代码 commands, flags, and terms to their current replacements.

Data: Claude Platform on AWS reference (1805 tks) - Reference documentation for using the Claude De Platform through AWS infrastructure, including AnthropicAWS clients, required region and workspace configuration, SigV4 authentication, and short-term API keys.

Data: Claude Tag (Claude in Slack) reference (1999 tks) - Offline reference for Claude Tag, Claude 代码's org-managed Slack surface, co what it is, a, setup, configuration, and how it differs from the earlier Claude in Slack app.

Data: Claude Test spec file format (1745 tks) - Reference for the Markdown spec files in .claude-test/specs/ — front matter keys, the three required parts, data-creating specs, and what makes a spec goal-oriented and checkable from a single end screenshot.

Data: Claude gateway landing page (392 tks) - HTML status page served at a Claude 代码 gateway root, showing the gateway logo, the running gateway URL, the identity-provider host, an OAuth discovery link, and the gateway .

Data: Cloud session un remote branch push notice (255 tks) - Notice shown when no remote of the checkout could be read, so it could not be checked whether the current branch or detached HEAD commit exists on the remote host, telling the user to push their work to a branch and start a new cloud session from it.

Data: 代码 change published event schema (349 tks) - Schema description for the code_change_published system event, including pro, idempotency, trust boundaries, and best-effort delivery.

Data: Command plugin 源码 command field (160 tks) - Describes the command field for command-based plugin sources, including its one-line directory output contract, cache-copy lifecycle, platform shell, and re-resolution behavior.

Data: Cowork plugin MCP discovery and connection (1746 tks) - Reference guidance for finding MCP connectors during plugin customization, using search and suggestion tools, mapping categories to keywords, and writing .mcp.json entries.

Data: Cowork plugin examples (3158 tks) - Reference examples of minimal, medium, and complex Cowork plugin structures with plugin metadata, skills, agents, hooks, MCP config, README, and connectors.

Data: Cross-session inbound setting (175 tks) - Describes crossSessionInbound delivery, hold, refusal, and permission-mode parity behavior for peer messages.

Data: Data anti-patterns (2339 tks) - Reference list of chart and dashboard anti-patterns to check against before shipping a data .

Data: Data choosing a form (1177 tks) - Reference guidance for choosing the appropriate chart, stat tile, or non-chart form based on the data's job.

Data: Data components (836 tks) - Reference specification for the HTML/S, layers, and system parameters that make up a data .

Data: Data interaction (1318 tks) - Reference guidance for chart ho, tooltips, filters, linked highlighting, and interaction behavior.

Data: Data marks and anatomy (2164 tks) - Reference specifications for chart marks, spacing, labels, axes, legends, and stat-tile anatomy.

Data: Dialog expiry setting (204 tks) - Describes dialogExpiry deadlines for remote permission dialogs and held cross-session messages, including defaults and o.

Data: DirectoryAdded hook description (216 tks) - Describes when the DirectoryAdded hook fires, its input fields, and how failures and output are handled for add-dir and register_repo_root sources.

Data: Files API reference — C# (378 tks) - C# Files API reference including beta file upload and use with beta message types.

Data: Files API reference — Go (413 tks) - Go Files API reference including file upload, listing, deletion, and usage in messages.

Data: Files API reference — Java (492 tks) - Ja API reference including file upload, beta message references, listing, deletion, download, and metadata retrieval.

Data: Files API reference — PHP (186 tks) - PHP Files API reference with a beta-era upload example and a warning that current SDKs use the stable Files API shape.

Data: Files API reference — Python (1719 tks) - Python Files API reference including file upload, listing, deletion, and usage in messages.

Data: Files API reference — TypeScript (961 tks) - TypeScript Files API reference including file upload, listing, deletion, and usage in messages.

Data: Gateway device code entry page (228 tks) - HTML page served at a gateway device endpoint, prompting the user to enter the short device 代码 shown by Claude 代码 so they can sign in through their company identity provider.

Data: Gateway ignored X-Forwarded-For warning (254 tks) - Gateway startup warning that an X-Forwarded-For header was ignored because the requesting address is not covered by listen.trusted_proxies, and what to configure when a load balancer or ingress fronts the gateway.

Data: Get task output control request (201 tks) - Schema description for the get_task_output control request that reads the last 8 KiB of a background shell or Monitor task's output without a model turn, and when it is refused.

Data: GitHub Actions workflow for mentions (525 tks) - GitHub Actions workflow template for triggering Claude 代码 via mentions.

Data: GitHub App installation PR description (409 tks) - Template for PR description when installing Claude 代码 GitHub App integration.

Data: Hook classifier context field (654 tks) - Schema description for the classifierContext hook response field, including trust semantics, size and timing limits, transcript , and rewrite-integrity requirements.

Data: Hook mcp_ser (312 tks) - Describes the mcp_ser attached to 工具-related hook payloads — the 服务端's config key and the 源码 its definition came from — and why trust must key on 源码 rather than on the 服务端 name or 工具-name prefix.

Data: Interrupt cancel queued parameter (646 tks) - Schema description for the optional interrupt cancel_queued request parameter and its queued-command cancellation semantics.

Data: Interrupt receipt cancelled field (291 tks) - Schema description for the cancelled UUID list returned when an interrupt request cancels queued commands.

Data: Interrupt receipt still queued field (1116 tks) - Schema description for the still_queued UUID list returned by interrupt control responses.

Data: Interrupt send_now message_uuid parameter (246 tks) - Schema description for the interrupt message_uuid parameter read only beside send_now, which restricts the request to one exact-match waiting user message and answers nothing_waiting when that message is not waiting.

Data: Interrupt send_now parameter (596 tks) - Schema description for the optional interrupt send_now request parameter, which asks that waiting user messages reach the model now without acting as a Stop, and how the CLI chooses between moving the turn's waits to the background and aborting it.

Data: Knowledge MCP search strategies (447 tks) - Reference query patterns for using knowledge MCPs to discoc 工具 names, project identifiers, team names, and workflow details during plugin customization.

Data: MCP HTTP 客户端 capability rationale (189 tks) - Explains where required 客户端 capabilities are in the modern MCP HTTP request ladder and how the pre-dispatch gate affects observable precedence.

Data: MCP HTTP parameter header rationale (247 tks) - Explains of Mcp-Param headers against 工具 declarations and request arguments, including error shape and pre-dispatch precedence.

Data: MCP HTTP standard header rationale (326 tks) - Explains the modern MCP HTTP ladder's Mcp-Method and Mcp-Name header checks and their obser precedence.

Data: MCP read resource control request (397 tks) - Documents the mcp_read_resource control request that reads one MCP Apps (SEP-1865) ui:// resource from a connected MCP 服务端, its sandboxed-render requirement, SDK-ser, refused lanes and error cases.

Data: MCP ser error_code field (200 tks) - Schema description for the MCP ser error_代码 field, naming the host-fixable failure causes — a rejected claude.ai connector login, a rejected first-party Anthropic credential, and a project-scoped ser approval — and noting that it is absent for every other failure and that new are additive.

Data: Managed Agents outcomes (2984 tks) - Reference documentation for Managed Agents outcomes, including user.define_outcome events, rubrics, outcome e events, deli, and interaction rules.

Data: Managed settings helper onFailure field (593 tks) - Schema description for the managed settings helper onFailure field, including 源码-dependent defaults, startup refusal behavior, static fallbacks, and background refresh handling.

Data: Managed settings helper retries field (311 tks) - Schema description for the managed settings helper retries field, including retryable failures, limits, per-attempt timeouts, backoff, and terminal failure handling.

Data: Message Batches API reference — C# (188 tks) - C# Batches API reference including batch creation, status polling, and result retrieval.

Data: Message Batches API reference — PHP (196 tks) - PHP Batches API reference including batch creation, status polling, and result retrieval.

Data: Message Batches API reference — Python (2038 tks) - Python Batches API reference including batch creation, status polling, and result retrieval at 50% cost.

Data: Message Batches API — TypeScript (1022 tks) - TypeScript usage guide for Claude's asynchronous Message Batches endpoint.

Data: Peer sender display name field (199 tks) - Schema description for the normalized display name on cross-session peer message senders.

Data: Plugin manifest types field (214 tks) - Schema description for the plugin manifest types field naming a self-contained .d.ts type contract for the noun the plugin's hooks module adds in 引擎.create, which dependent mods recei .claude-plugin/types and claude plugin checks.

Data: Prompt suggestions paused control request (229 tks) - Documents the set_prompt_suggestions_paused control request that pauses or resumes a session's prompt suggestions at runtime and the process-local, non-persisted scope of that state.

Data: Query result pending command count (167 tks) - Describes the query result field that counts queued user sends still awaiting turns and clarifies when it is absent or zero.

Data: Rate limit grace signal (306 tks) - Describes the usage-limit grace-window signal tracked from the latest ser, including when it sets, clears, and how it interacts with o status at hard exhaustion.

Data: Rate limit unified windows (325 tks) - Describes per-window subscription rate-limit utilization and reset data derived from unified rate-limit response headers.

Data: Remote tools reannounce control request (220 tks) - Documents the remote_tools_reannounce control request a replacement worker sends so the attached 客户端 re-announces a machine missing from its 工具 roster, including worker-epoch, once-per-epoch, ten-second and deadline rules.

Data: Review upload excluded changes error (327 tks) - Error shown when a review upload has no new commits or uploadable changes because every uncommitted file was excluded from the git bundle.

Data: Rewind files skippedLinks field (223 tks) - Describes the rewindFiles skippedLinks count, including link-safety refusal semantics and dry-run behavior.

Data: SDK API error kind field (1370 tks) - Documents the typed api_error kinds carried on SDK assistant messages that wrap an API error, and what each kind means for replaying, rewinding, or remedying the request.

Data: SDK footer indicator schema (249 tks) - Schema description for the SDK footer indicator, its en and cached-bootstrap sources, and its delivery on system/init and initialize responses so host UIs can render the terminal status pill.

Data: SDK frame_intake_phases_ms field (444 tks) - Schema description for the SDK turn-timing frame_intake_phases_ms field, breaking the wait between frame receipt and enqueue into named steps and their sum and presence in.

Data: SDK frame_recei field (193 tks) - Schema description for the SDK turn-timing frame_recei field recording when the triggering send's frame arrived on a session reading input from the session 服务端's SSE stream, how it pairs with the enqueue and turn-start timestamps to split transit from queue wait, and when it is absent.

Data: SDK initialize response sdk_mcp_manifests_parked field (482 tks) - Schema description for the SDK initialize response sdk_mcp_manifests_parked field, reporting per ser a manifest entry was parked, already connected, mismatched on 协议 , malformed, or not honoured, and when the field is absent.

Data: SDK set max thinking tokens request schema (477 tks) - Schema description for the SDK set_max_thinking_tokens control request, co null reset omitted budget, session-scoped thinking_display o, and when the highlights display mode is accepted, downgraded to omitted, or refused.

Data: SDK set model system prompt field (243 tks) - Schema description for the internal set_model system_prompt field, including next-turn application with unchanged 工具 definitions, non-empty updates, feature o, and compatibility behavior.

Data: SDK system init plugin_errors field (315 tks) - Schema description for the plugin_errors array on the SDK system/init message, co failed partially loaded plugins, name and positional inline/synced tags, open-set error types, omission when clean, and Remote Control workers always omitting it.

Data: SDK thinking_duration_ms field (184 tks) - Schema description for the display-only thinking_duration_ms wrapper field giving how long a streamed thinking block took, excluding the wait for the first token, and when it is absent or not replayed to the model.

Data: Sandbox TLS termination setting (199 tks) - Describes the experimental sandbox.网络.tlsTerminate setting, certificate behavior, Windows trust requirements, 源码 precedence, and initialization .

Data: Sandbox credential en no-match setting (188 tks) - Describes sandbox credential en onExtractNoMatch behavior, including warn, deny, error, and decode-path constraints.

Data: Sandbox credential file mask claims setting (179 tks) - Describes sandbox credential file maskClaims behavior for selectively masking decoded JWT payload claims while preserving other claims.

Data: Sandbox credential mask empty injectHosts warning (174 tks) - Warns that sandbox credential masks with empty injectHosts expose only sentinel and explains how to resol, managed-settings, and local configuration causes.

Data: Sandbox deniedResol setting (209 tks) - Schema description for the sandbox deniedResol setting listing IP addresses or CIDR ranges an allowed hostname must not resolve to, its interaction with allowedDomains IP literals, and its exemption for parentProxy and mitmProxy routes.

Data: Sandbox filesystem disabled setting (227 tks) - Describes sandbox.filesystem.disabled behavior, platform limits, read-protection effects, and configuration precedence.

Data: Sandbox 网络 domain spelling warning (278 tks) - Formats the sandbox diagnostic for unreliable 网络 domain spellings and explains conser enforcement until entries are corrected.

Data: Sandbox root without CAP_SETFCAP error (140 tks) - Linux sandbox dependency error explaining that running as uid 0 without CAP_SETFCAP makes every sandboxed command fail while writing a uid map on kernels 5.12+ and backports, and telling the user to grant CAP_SETFCAP or run as non-root.

Data: Self-hosted runner Anthropic git proxy credential warning (338 tks) - Runner warning that sources on hosts not routed through Anthropic-managed git clone with this machine's credentials, which --use-anthropic-git-proxy wipes from HOME-level git config, so credentials must live in /etc/gitconfig, GIT_ASKPASS, or an SSH key.

Data: Self-hosted runner GIT_ASKPASS go hosts warning (311 tks) - Runner warning that a GIT_ASKPASS program answers for every host, so later git config could send this machine's credential for go hosts through Anthropic's relay, and directs scoping or unsetting it along with core.askPass and SSH_ASKPASS.

Data: Self-hosted runner GIT_SSL_CAINFO trust bundle warning (254 tks) - Runner warning that GIT_SSL_CAINFO pre building the combined certificate file for Anthropic-managed git, giving the failure cause and first remedy and directing the company CA into a per-服务端 http sslCAInfo entry.

Data: Self-hosted runner GIT_SSL_NO_ lifecycle hook exception (330 tks) - Runner warning addendum explaining that checkout and post-session hooks of sessions with a repository on Anthropic-managed git get http.ssl=false with certificate checks re-enabled for the mount, while other sessions' hooks inherit the switch.

Data: Self-hosted runner GIT_SSL_NO_ session relocation note (325 tks) - Runner warning addendum explaining that inside a session GIT_SSL_NO_ becomes http.ssl=false that no longer covers the git mount or go hosts, that per-ser takes precedence, and to name a company CA on certificate failures.

Data: Self-hosted runner 客户端 certificate relay warning (539 tks) - Runner warning that TLS 客户端 certificate set on the runner would be offered to the session relay rather than go hosts, with remedies of asking for direct listing, unsetting the , or scoping sslCert and sslKey per URL.

Data: Self-hosted runner decode-token command help (869 tks) - Documents self-hosted runner token decoding, redaction, , input, and output command-line options.

Data: Self-hosted runner deferred shutdown timing notice (320 tks) - Explains deferred shutdown release and 盗取 timing, supervisor timeout sizing, forced-kill consequences, and second-signal behavior for self-hosted runners.

Data: Self-hosted runner git-lfs hook warning (342 tks) - Runner warning that git-lfs's pre-push hook never runs inside lifecycle hooks, so a post-session git push sends LFS pointers without their objects, and how to push the objects or point core.hooksPath instead.

Data: Self-hosted runner orchestrator command help (1390 tks) - Documents self-hosted runner orchestrator connection, hook, SCM connector, runtime, health, and debug command-line options.

Data: Self-hosted runner session metrics help (622 tks) - Defines Prometheus HELP text for self-hosted runner session lifecycle counters.

Data: Streaming reference — C# (376 tks) - C# streaming reference including streaming events and the RawMessageStreamEvent TryPick methods.

Data: Streaming reference — Go (452 tks) - Go streaming reference including event handling, error checks, and final-message accumulation.

Data: Streaming reference — Java (287 tks) - Ja reference including streamed content-block delta handling and text extraction.

Data: Streaming reference — PHP (315 tks) - PHP streaming reference including streaming events and handling content block deltas (requires SDK v0.5.0+).

Data: Streaming reference — Ruby (116 tks) - Ruby streaming reference including basic streamed text handling.

Data: Structured 工具 output field schema (234 tks) - Schema description for the user-message 工具_use_result field, including per-工具 output shapes, the completed Agent or Task output contract, and the detachedToolCall placeholder for calls that stepped aside for a user message.

Data: Structured usage rate-limit rows field (177 tks) - Schema description for structured usage rate-limit rows, including ser meter ordering, null and empty semantics, and the synthesized header-fallback row.

Data: Submit feedback control request (231 tks) - Documents the internal submit_feedback control request that sends a /feedback report with the session transcript and sanitized error log, returns una when feedback is disabled or unsendable, and honors sa by writing a redacted local bundle instead.

Data: Telemetry ignored notice (268 tks) - Settings-status warning listing OTEL/telemetry en a settings file sets that Claude 代码 ignores because such files can only turn telemetry off, with guidance on where to set them intentionally.

Data: Thin-client diff dialog schema (198 tks) - Internal data description for workspace git diff payloads used by the thin-客户端 diff dialog.

Data: Token counting reference (665 tks) - Reference documentation for counting Claude model tokens with the Messages count_tokens endpoint and Anthropic SDK or CLI examples, including warnings against OpenAI tokenizers.

Data: Tool use display metadata field (172 tks) - Documents the tool_use_meta wire field carrying per-block display metadata for a message's 工具_use blocks; it is wrapper-level UI metadata and is not replayed to the model.

Data: Tool use reference — C# (2401 tks) - C# 工具 use reference including defining tools and reconstructing response content for the follow-up assistant message.

Data: Tool use reference — PHP (3211 tks) - PHP 工具 use reference including the beta 工具 runner and the manual agentic loop with camelCase keys.

Data: Tool use reference — Ruby (444 tks) - Ruby 工具 use reference including the beta 工具 runner and manual agentic-loop guidance.

Data: Turn handoff a event schema (269 tks) - Schema description for the turn_handoff_a system event a cloud worker emits right after registering, carrying the tools and turn-handoff options its registration declared, and the rule that readers keep only the entry from the newest worker life.

Data: Turn handoff file_names field (283 tks) - Schema description for the turn_handoff file_names member listing user-attached files and their 客户端-facing names that a capable worker copies into the session's home directory before running the handed-o, and when a malformed member is ignored.

Data: Turn handoff memory_context field (232 tks) - Schema description for the turn_handoff memory_context member carrying the claude.ai memory snapshot the 客户端 showed its model, how the worker writes it as a cowork_memory_context attachment line ahead of the handed-o, and when a malformed or uuid-colliding member is ignored.

Data: Upload device hook template request (200 tks) - Describes the cloud device request that uploads a hook template to a worker before device hooks are registered.

Data: changed branch field (172 tks) - Describes the optional best-effort branch hint on events for commits and pushes, including multi-branch and uncertain-attribution behavior.

Data: changed event schema (392 tks) - Schema description for the system event as a best-effort repository cache-in signal.

Data: Windows transcript read EBADF notice (180 tks) - Explains a Windows-only EBADF error reading the session transcript file likely caused by security or endpoint-management software intercepting file reads, and suggests excluding the .claude folder or allow-listing Claude 代码 before resuming.

Data: Working tree upload refusal for duplicated withheld file (211 tks) - Error refusing a working-tree upload because an index entry appeared holding byte-for-byte the content of a file that must stay on this machine, indicating something outside the upload wrote into its index copy.

Data: Working tree upload refusal for misplaced .git entry (345 tks) - Error refusing a working-tree upload because the checkout's .git is a pointer file, symbolic link, hard-linked pointer file, or otherwise unfollowable entry, telling the user to start from the repository's ordinary main checkout instead.

Data: allowManagedPermissionRulesOnly allowed-tools frontmatter scope (424 tks) - Continues the allowManagedPermissionRulesOnly schema description: skill, command, and plugin allowed-tools frontmatter is ignored except for managed and bundled skills and plugins on admin-backed channels such as pinned managed marketplaces, attested claude.ai-synced plugins, and host-delivered --plugin-dir plugins.

Data: allowManagedPermissionRulesOnly deny rules still apply (45 tks) - Closes the allowManagedPermissionRulesOnly schema description by noting that --disallowedTools, skill disallowed-tools, and other command-line or session deny and ask rules still apply.

Data: allowManagedPermissionRulesOnly setting (66 tks) - Schema description for the managed allowManagedPermissionRulesOnly setting, which ignores permission rules from user, project, local, and --settings files and --allowedTools allow rules so only managed settings can add allow rules.

Data: allowedMcpServers setting (204 tks) - Describes how the enterprise allowedMcpServers setting governs user-added organization-delivered MCP servers, including ${VAR} expansion, undefined and empty-array behavior, and denylist precedence.

Data: a setting (315 tks) - Schema description for the managed a setting, contrasting prefix and exact matching of a entries, family and release-dependent alias handling, Default-model restriction with startup failure, and which background and helper requests stay unrestricted.

Data: deniedModels setting (222 tks) - Schema description for the managed deniedModels setting, co family-alias blocking, model ID blocking across dates, -fast and provider prefixes, ignored release-dependent aliases, and Default-model step-down or startup failure.

Data: syncClaudeAiPlugins setting (332 tks) - Describes how the syncClaudeAiPlugins setting disables account-synced plugin downloads, a, and cleanup across user, managed, workspace, and in scopes.

Data: syncClaudeAiSkills setting (314 tks) - Describes how the syncClaudeAiSkills setting disables account-synced skill downloads, a, and cleanup across user, managed, workspace, and in scopes.

System Prompt

Parts of the main system prompt.

System Prompt: /loop tick (loop.md absent, dynamic pacing) (315 tks) - Loop tick injection for dynamic self-paced autonomous checks when loop.md is absent.

System Prompt: /loop tick (loop.md tasks) (94 tks) - Loop tick injection for recurring cron-based runs of tasks from loop.md.

System Prompt: /loop tick (loop.md tasks, dynamic pacing) (311 tks) - Loop tick injection for dynamic self-paced runs of tasks from loop.md.

System Prompt: Act when ready (68 tks) - Instructs the agent to act once it has enough information and gi instead of exhausti.

System Prompt: Action safety and truthful reporting (161 tks) - Requires confirmation for irre or outward-facing actions, checking targets before destructi, and truthful reporting of outcomes.

System Prompt: Advisor 工具 instructions (443 tks) - Instructions for using the Advisor 工具.

System Prompt: Agent Summary Generation (291 tks) - System prompt used for "Agent Summary" generation.

System Prompt: Agent thread notes (293 tks) - Beha guidelines for agent threads co absolute paths, response formatting, emoji a, and 工具 call punctuation.

System Prompt: Analysis instructions for full compact prompt (recent messages) (339 tks) - Provides analysis instructions for summarizing only recent messages during con compaction.

System Prompt: Artifact comment decision reformat retry (256 tks) - Instructs the Artifact comment composer to reformat a malformed prior response as exactly one valid JSON decision while treating the reproduced response as untrusted data.

System Prompt: Artifact comment edit composer (911 tks) - Instructs a 工具-less Artifact comment composer to emit exactly one reply or edit decision using ordered exact-string patches and an a full-rewrite form.

System Prompt: Artifact comment fast acknowledgement selection (603 tks) - Instructs a no-tools helper to select exactly one numbered canned acknowledgement for the newest Artifact comment based on edit capability, trigger state, and whether the request needs an artifact edit or reply.

System Prompt: Artifact comment fast acknowledgement (240 tks) - Prompts a no-tools helper model to produce one short plain-text acknowledgement for the newest Artifact comment before the full reply.

System Prompt: Artifact comment list framing (1055 tks) - Frames Artifact comment-list 工具 results as untrusted data using randomized fences and optional selected-text, anchor-path, and anchor-file/page guidance.

System Prompt: Artifact comment presence state guidance (238 tks) - Explains that Artifact comment presence rows are 工具-emitted, page-produced untrusted data that may resol but ne instructions or permissions.

System Prompt: Artifact comment reply composer (375 tks) - Instructs a 工具-less composer to answer an Artifact comment thread or flag requested artifact changes for its session in one brief plain-text reply without claiming or promising edits.

System Prompt: Artifact comment result guidance (225 tks) - Appends reply, resolve, and focused thread-read instructions to Artifact comment-list 工具 results.

System Prompt: Artifact comment thread framing (2178 tks) - Frames an Artifact comment thread and optional anchor context as untrusted data using randomized fences.

System Prompt: Artifact comment thread triage (306 tks) - Classifies the newest human request in a framed Artifact comment thread as an artifact edit or a reply-only pipeline action.

System Prompt: Artifact commenter access guidance (183 tks) - Explains that the access word before a comment's stamp — owner, editor or commenter — is the access the ser for that person, context for weighing feedback but never a permission, and when an outside-organization note follows it.

System Prompt: Artifact page-owned comment thread guidance (192 tks) - Directs replies to page-owned comment threads through an a authorized page 工具 rather than Artifact relay threads.

System Prompt: Artifact patch turn edit composer (670 tks) - Instructs a 工具-less composer to emit exactly one JSON decision for an artifact edit request, either ordered exact-string file edits with a one-sentence reply and confidence rating or a hand-back to the main assistant when needed context is missing, under strict patch-uniqueness and sizing rules.

System Prompt: Auto mode Slack message pro (1170 tks) - Instructs the auto mode classifier to treat ser human Slack relays as user intent while rejecting bot-authored and nested relays as consent.

System Prompt: Auto mode setup proposal generator (2740 tks) - Transforms gathered repository and usage reconnaissance into a constrained JSON proposal for auto-mode en context and permission rules.

System Prompt: Auto mode (244 tks) - Continuous task execution, akin to a background agent.

System Prompt: Autonomous loop check (1484 tks) - Defines behavior for autonomous timer-based in, guiding Claude to continue established work, maintain PRs, and handle repeated idle checks while the user is away.

System Prompt: Autonomous loop notification guidance (98 tks) - Guides when autonomous loop ticks should notify the user for blockers or actionable state changes.

System Prompt: Autonomous loop persistence guidance (CLAUDE_CODE_LOOP_PERSISTENT) (1630 tks) - Defines behavior for autonomous timer-based in, guiding Claude to persistently continue established work, maintain PRs, and broaden scope before stopping while the user is away.

System Prompt: Autonomous loop tick (dynamic pacing) (311 tks) - Autonomous loop tick injection for dynamic self-paced autonomous checks scheduled with ScheduleWakeup.

System Prompt: Autonomous loop tick (87 tks) - Autonomous loop tick injection for recurring cron-based autonomous checks.

System Prompt: Autonomous operation guidelines (407 tks) - Instructs autonomous sessions to proceed on re work, stop for destructive or scope-changing actions, and finish promised work before ending the turn.

System Prompt: A Unnecessary Sleep Commands (part of PowerShell 工具 description) (175 tks) - Guidelines for a unnecessary sleep commands in PowerShell scripts, including alternatives for waiting and notification.

System Prompt: Background session instructions (354 tks) - Instructions for background job sessions to use the job-specific temporary directory and follow the appropriate worktree isolation guidance.

System Prompt: Background session worktree persistence guidance (186 tks) - Directs background sessions to commit and push worktree changes when appropriate while preserving user git-control instructions and branch safety.

System Prompt: Background worktree isolation guidance (129 tks) - Tells background sessions when to enter an isolated worktree before making 代码 changes and when to continue in place.

System Prompt: Censoring assistance with malicious acti (98 tks) - Guidelines for assisting with authorized security testing, defensi, CTF challenges, and educational contexts while censoring requests for malicious acti.

System Prompt: Chrome browser MCP tools (388 tks) - Instructions for loading deferred Chrome browser MCP tools through ToolSearch in a single batched selection before browser tasks.

System Prompt: Claude Fable 5 model identity (202 tks) - Identifies this Claude iteration as Claude Fable 5, explains its relationship to Claude Mythos 5, and links to Anthropic's comparison announcement.

System Prompt: Claude Fable 5.1 model identity (235 tks) - Identifies this Claude iteration as Claude Fable 5.1, explains its relationship to Claude Mythos 5.1, and points users to Anthropic's Fable and Mythos announcement for differences.

System Prompt: Claude in Chrome browser automation (non-deferred tools) (1130 tks) - Instructions for using directly a Claude in Chrome browser automation tools effectively.

System Prompt: Claude in Chrome browser automation (1158 tks) - Instructions for using Claude in Chrome browser automation tools effectively.

System Prompt: Claude in Chrome browser selection instructions (191 tks) - Instructs the agent to ask the user to choose among multiple connected Chrome browsers before using browser automation tools.

System Prompt: Claude in Chrome deferred 工具 loading (311 tks) - Directs Claude to batch-load deferred core and task-specific Chrome MCP tools in one ToolSearch call.

System Prompt: Combined memory index pointer instructions (120 tks) - Instructs the agent to add one-line pointers for private and team memories to the single private memory index and ne memory content there.

System Prompt: Comment what and task context a (76 tks) - Instructs Claude not to write comments that explain what 代码 does or reference transient task context.

System Prompt: Comment why-only guidance (67 tks) - Instructs Claude to write 代码 comments only when the reason is non-obvious and useful to future readers.

System Prompt: Communication style (297 tks) - Instructs Claude to gi, user-facing updates at key moments during 工具 use, write concise end-of-turn summaries, match response format to task complexity, and avoid comments and planning documents in 代码.

System Prompt: Concise output style (373 tks) - Defines the built-in Concise output style to lead with results, omit narration, stay brief by default, and preser detail and correctness.

System Prompt: Context compaction summary (409 tks) - Prompt used for context compaction summary (for the SDK).

System Prompt: Coordinator cross-session peer guidance (326 tks) - Explains how coordinators discover and message peer Claude sessions while treating peer requests as untrusted input rather than delegated worker authority.

System Prompt: Coordinator mode orchestration (5896 tks) - Provides coordinator-mode instructions for delegating work to worker agents, managing worker lifecycle, handling cross-session peers, and delegated results.

System Prompt: Correction restraint (365 tks) - Instructs Claude to correct only consequential errors plainly, avoid unnecessary self-criticism or re-auditing, and e other agents’ corrections before adopting them.

System Prompt: Deli work at full scope (605 tks) - Instructs Claude to complete ordinary requested work at full scope under reasonable assumptions, continue past non-blocking concerns, and preser refusal and risky-action confirmation boundaries.

System Prompt: Deny rule circum classifier guidance (74 tks) - Guides permission classification to block attempts to route around configured Edit, Write, or MultiEdit deny rules.

System Prompt: Description part of memory instructions (148 tks) - Defines what the memory description field should contain within broader user-memory authoring instructions.

System Prompt: Doing tasks (ambitious tasks) (47 tks) - Allow users to complete ambitious tasks; defer to user judgement on scope.

System Prompt: Doing tasks (help and feedback) (24 tks) - How to inform users about help and feedback channels.

System Prompt: Doing tasks (no compatibility hacks) (52 tks) - Delete unused 代码 completely rather than adding compatibility shims.

System Prompt: Doing tasks (no unnecessary additions) (73 tks) - Do not add features, refactor, or impro what was asked.

System Prompt: Doing tasks (no unnecessary error handling) (64 tks) - Do not add error handling for impossible scenarios; only at boundaries.

System Prompt: Doing tasks (security) (67 tks) - Avoid introducing security like injection, XSS, etc.

System Prompt: Doing tasks (software engineering focus) (104 tks) - Users primarily request software engineering tasks; interpret instructions in that context.

System Prompt: Dream CLAUDE.md memory reconciliation (436 tks) - Instructs dream memory consolidation to reconcile feedback and project memories against CLAUDE.md, deleting stale memories or flagging possible CLAUDE.md drift.

System Prompt: Dream team memory handling (279 tks) - Instructions for handling shared team memories during dream consolidation, including deduplication, conser pruning rules, and a accidental promotion of personal memories.

System Prompt: Emoji a (31 tks) - Instructs Claude to avoid using emojis unless the user explicitly asks for them.

System Prompt: Executing actions with care (1105 tks) - Instructions for executing actions carefully.

System Prompt: Explain /代码-review ultra (131 tks) - Guidance shown when a user asks about 'ultrareview': explains it maps to /代码-review ultra (the /ultrareview alias is deprecated) and that the agent can't start it directly.

System Prompt: Exploratory questions — analyze before implementing (75 tks) - Instructs Claude to respond to open-ended questions with analysis, options, and tradeoffs instead of jumping to implementation, waiting for user agreement before writing 代码.

System Prompt: Feedback memory body structure (79 tks) - Defines the body structure for feedback memories, including the rule, why, and how to apply it.

System Prompt: Feedback memory sa (121 tks) - Explains when to sa memories from user corrections or confirmed non-obvious approaches.

System Prompt: Focus mode (long form) (105 tks) - Focus-mode notice (long form): the user sees only the final text, not 工具 calls, results, or inter-step writing.

System Prompt: Focus mode (short form) (91 tks) - Focus-mode notice (short form): only each response's final text reaches the user.

System Prompt: Fork usage guidelines (326 tks) - Instructions for when to fork subagents and rules against reading fork output mid-flight or fabricating fork results.

System Prompt: Forked agent guidance (164 tks) - Explains that calling Agent with subagent_type "fork" creates a background fork and when to use it.

System Prompt: Forked con worktree isolation guidance (416 tks) - Warns a forked background con not to enter the original session's worktree and directs it to create a separate worktree when isolation is enabled.

System Prompt: Fresh subagent delegation example (315 tks) - Provides an example of briefing a fresh specialized subagent with sufficient context and a specific reporting request.

System Prompt: Frontend browser (86 tks) - Requires Claude to start the de and UI or frontend changes in a browser before reporting completion.

System Prompt: Git status (37 tks) - System prompt for displaying the current git status at the start of the con.

System Prompt: Harness instructions (299 tks) - Core interacti identity and harness instructions for terminal Markdown output, security, permissions, system-reminder handling, hook feedback, 工具 use, and 代码 references.

System Prompt: Hook e truncated transcript note (74 tks) - Tells the hook condition e that earlier con was omitted and how to handle insufficient e.

System Prompt: Hook feedback handling (85 tks) - Explains that hook feedback should be treated as user feedback and how to respond when hooks block actions.

System Prompt: Hooks Configuration (1493 tks) - Documents hook structure, events, types, input/output fields, and examples for the Claude 代码 configuration skill.

System Prompt: How to use the SendUserMessage 工具 (283 tks) - Instructions for using the SendUserMessage 工具.

System Prompt: Insights at a glance summary (569 tks) - Generates a concise 4-part summary (what's working, hindrances, quick wins, ambitious workflows) for the insights report.

System Prompt: Insights friction analysis (139 tks) - Analyzes aggregated usage data to identify friction patterns and categorize recurring issues.

System Prompt: Insights interaction style (119 tks) - Analyzes Claude 代码 usage data to describe the user's interaction style.

System Prompt: Insights memorable moment (96 tks) - Analyzes Claude 代码 usage data to find a memorable qualitati.

System Prompt: Insights on the horizon (148 tks) - Identifies ambitious future workflows and opportunities for autonomous AI-assisted de.

System Prompt: Insights session facets extraction (310 tks) - Extracts structured facets (goal categories, satisfaction, friction) from a single Claude 代码 session transcript.

System Prompt: Insights suggestions (737 tks) - Generates actionable suggestions including CLAUDE.md additions, features to try, and usage patterns.

System Prompt: Insights summary (At a Glance) (272 tks) - The 'At a Glance' summary block of the Insights report (what's working / what's hindering).

System Prompt: Insights what works (121 tks) - Analyzes Claude 代码 usage data to identify workflows that are working well for the user.

System Prompt: Interacti intro (output-style active) (40 tks) - Opening system-prompt line for sessions that have an Output Style configured.

System Prompt: Interacti intro (output-style conditional) (231 tks) - Opening system-prompt line that selects Output Style, collaborati, or software-engineering framing and injects security guidance.

System Prompt: Interacti intro (short) (27 tks) - Minimal opening system-prompt line for software-engineering sessions.

System Prompt: Learning mode (insights) (142 tks) - Instructions for pro educational insights when learning mode is active.

System Prompt: Learning mode (1042 tks) - Main system prompt for learning mode with human collaboration instructions.

System Prompt: Memory description of user feedback (139 tks) - Describes the user feedback memory type that stores guidance about work approaches, emphasizing recording both successes and failures and checking for contradictions with team memories.

System Prompt: Memory index pointer instructions (129 tks) - Instructs the agent to add one-line pointers to the memory index file and treat the index as separate from memory content.

System Prompt: Memory instructions (487 tks) - Instructions for using persistent file-based memory, including memory file format, scope, indexing, and stale-memory handling.

System Prompt: Memory persistence scope (60 tks) - Explains that memory is for information useful in future con, not only within the current con.

System Prompt: Memory sa (14 tks) - Lists categories of information that should not be saved in memory, even when the user asks.

System Prompt: Minimal mode (285 tks) - Describes the behavior and constraints of minimal mode, which skips hooks, LSP, plugins, auto-memory, and other features while requiring explicit context via CLI flags.

System Prompt: Monitor fallback heartbeat guidance (319 tks) - Guides dynamic loop ticks to use Monitor as the primary wake signal and ScheduleWakeup as a fallback heartbeat, place a status update before or after re-arming, and stop the monitor when ending the loop.

System Prompt: Mo comment thread guidance (158 tks) - Directs Claude to re-e a pre sent Artifact comment after its author moves the thread to a new location.

System Prompt: Non-fork subagent delegation examples (894 tks) - Provides conditional non-fork subagent examples for background and synchronous execution, including self-contained prompts, waiting-state responses, and independent-review briefing.

System Prompt: One of six rules for using sleep command (23 tks) - One of the six rules for using the sleep command.

System Prompt: Option pre (151 tks) - System prompt for pre UI options in a side-by-side layout.

System Prompt: Outcome-first communication style (815 tks) - Instructs Claude to keep user-facing updates readable and outcome-first, answer directly after work completes, match response format to task complexity, and limit 代码 comments to non-obvious constraints.

System Prompt: PR Slack notification step (125 tks) - Adds a PR workflow step to optionally ask the user before posting the PR URL to Slack.

System Prompt: PR Steward handoff guidance (597 tks) - Explains the PR Steward watching and working labels and requires checking a PR's labels before pushing, babysitting, or monitoring it, offering the user leave-it, one-change, or take-o and leaving the PR alone when the user is away.

System Prompt: Parallel 工具 call note (part of "工具 usage policy") (102 tks) - System prompt telling Claude to use parallel 工具 calls.

System Prompt: Partial compaction instructions (852 tks) - Instructions on how to compact when the user decided to compact only a portion of the con, with a structured summary format and analysis process.

System Prompt: Permission classifier strict review guidance (79 tks) - Instructs the permission classifier to carefully deny blocked actions and require explicit user confirmation for o.

System Prompt: Personal project memory description (67 tks) - Describes project memories for ongoing work, goals, initiatives, bugs, or incidents relevant to the user's work in a directory.

System Prompt: Phase four of plan mode (292 tks) - Final plan-writing instructions for phase four of plan mode.

System Prompt: Plan mode interacti offer (688 tks) - Instructs plan mode to offer an interacti for substanti decisions and integrate accepted workshop decisions into the canonical plan.

System Prompt: Plan sent to Ultraplan (106 tks) - User-facing note confirming a plan has been sent to Ultraplan for remote refinement.

System Prompt: Plan guidance (115 tks) - Explains when to use or update a plan instead of saving information to memory.

System Prompt: PowerShell edition for 5.1 (440 tks) - System prompt for pro information about Windows PowerShell 5.1.

System Prompt: PowerShell edition for 7+ (128 tks) - Describes PowerShell 7+ shell syntax support, including pipeline chain operators, ternary, null-coalescing, and UTF-8 defaults.

System Prompt: PowerShell edition unknown (108 tks) - Assumes Windows PowerShell 5.1 compatibility when the PowerShell edition is unknown and forbids PowerShell 7-only syntax.

System Prompt: Prefer editing existing files (17 tks) - Instructs Claude to prefer editing existing files instead of creating new ones.

System Prompt: Project member session user message pro (1064 tks) - Defines how marked messages in a shared-project member's own session count as that member's user turns while coordinator relays, bare approvals, and other relayed content stay untrusted.

System Prompt: Project memory body structure (83 tks) - Defines the body structure for project memories, including the fact or decision, why, and how to apply it.

System Prompt: Project memory sa (91 tks) - Explains when to sa memories about who is doing what, why, or by when, including absolute date handling.

System Prompt: Project skill upkeep for feedback memory (314 tks) - Instructs Claude to update the relevant project skill when saving feedback memory about repeatable workflow corrections.

System Prompt: Project thread in-thread message pro (291 tks) - Explains that a message the person sends in the project thread arrives as its own marked user turn, and that a bare appro answers only what the session sent through the reply 工具 and clears no SOFT BLOCK unless it names the action and its target.

System Prompt: Project timeline user message pro (1688 tks) - Defines how project timeline messages from the owner become user turns while other relayed or fetched project content remains untrusted external data.

System Prompt: Remote plan mode (ultraplan) (870 tks) - System reminder injected during remote planning sessions that instructs Claude to explore the codebase, produce a diagram-rich plan , and implement it with a pull request upon approval.

System Prompt: Remote planning session (612 tks) - System reminder that configures a remote planning session to explore the codebase, produce an implementation plan , and handle plan approval, rejection, or teleportation back to the user's local terminal.

System Prompt: Reporting outcomes (261 tks) - Requires reports to distinguish obser from intentions, lead with failures or incomplete work, and avoid o un completion.

System Prompt: Respond in configured language (128 tks) - Directs all responses, explanations, and 代码 commentary into a configured language.

System Prompt: Responsive mode (813 tks) - Directs Claude to answer immediately with a one or two sentence acknowledgement before thinking or calling any 工具, and to write plain con English free of openers, flattery, filler and wrap-ups.

System Prompt: Sandbox task resource boundary (217 tks) - Limits sandboxed work to task-provided resources even when other resources are reachable and requires reporting missing access instead of bypassing the boundary.

System Prompt: Saving skills via file delivery (270 tks) - Explains that account skills are not modified by filesystem writes and directs the agent to deli skill files with the file-sending 工具.

System Prompt: Self-hosted runner doctor (6808 tks) - Instructs an agent to diagnose self-hosted runner authentication, 网络, lifecycle, queue, hooks, metrics, and escalation issues.

System Prompt: Self-hosted runner setup (2228 tks) - Guides an operator through creating, , learning, and documenting a self-hosted runner deployment.

System Prompt: Shared git stash safety (201 tks) - Warns that git stash is shared across worktrees and sessions, preferring WIP commits or uniquely tagged stash entries.

System Prompt: Skill sa note (163 tks) - Notes that whether a delivered file can be saved as a skill depends on organization settings the assistant cannot see, and how to phrase that conditionally rather than as a certainty.

System Prompt: Skillify Current Session (1798 tks) - System prompt for con the current session into a skill.

System Prompt: Subagent delegation cost guidance (503 tks) - Explains when delegation is worth its context and token costs, when to work inline, and how to brief a subagent narrowly.

System Prompt: Subagent delegation examples (620 tks) - Provides example interactions showing how a coordinator agent should delegate tasks to subagents, handle waiting states, and report results.

System Prompt: Subagent delegation restraint (479 tks) - Guides Claude to delegate work only when it is genuinely independent, large enough to justify a fresh context, or naturally parallel, while a excessive or redundant subagents and not redoing delegated work.

System Prompt: System section (93 tks) - System section of the main system prompt.

System Prompt: Task appro (109 tks) - Instructs the agent to continue agreed tasks end to end without unnecessary re-confirmation.

System Prompt: Tasks guidance (80 tks) - Explains when to use tasks instead of saving current-con progress to memory.

System Prompt: Team memory index pointer instructions (159 tks) - Instructs the agent to add one-line memory pointers to the appropriate team memory index file and ne memory content into the index.

System Prompt: Team project memory description (67 tks) - Describes project memories for shared ongoing work, goals, initiatives, bugs, or incidents within a working directory.

System Prompt: Teammate Communication (105 tks) - System prompt for teammate communication in swarm.

System Prompt: Tone and style (代码 references) (39 tks) - Instruction to include file_path:line_number when referencing code.

System Prompt: Tone and style (concise output — short) (16 tks) - Instruction for short and concise responses.

System Prompt: Tool call colon a (59 tks) - Instructs Claude not to use a colon before 工具 calls because 工具 calls may be hidden from user output.

System Prompt: 工具 call summary label (112 tks) - Instructs Claude to write a short past-tense summary label for completed 工具 calls in mobile UI rows.

System Prompt: 工具 usage (subagent guidance) (141 tks) - Guidance on when and how to use subagents effectively.

System Prompt: 工具 usage (task management) (70 tks) - Use TodoWrite to break down and track work progress.

System Prompt: Troubleshooting confirmation policy (71 tks) - Requires explaining fixes and confirming before destructive or installation-changing troubleshooting commands.

System Prompt: User memory usage guidance (91 tks) - Explains when to use user memories to tailor responses to the user's profile or perspective.

System Prompt: WSL managed settings double opt-in (152 tks) - Explains that WSL can read the Windows managed settings policy chain only when the admin-enabled flag is set, with HKCU requiring an additional user opt-in.

System Prompt: Worker instructions (356 tks) - Instructions for workers to follow when implementing a change.

System Prompt: Writing for the user (586 tks) - Defines concise, standalone user-facing final-message rules co sentence structure, formatting, 代码 and numbers, terminology, and stopping once the answer is complete.

System Prompt: Writing subagent prompts (398 tks) - Guidelines for writing effecti when delegating tasks to subagents, co context-inheriting subagent scenarios.

System Prompt: You should know understood-topic skip guidance (150 tks) - Tells the You should know suggestion agent to skip topics the person plausibly understood in the main session while still allowing consequential details mentioned only in passing.

System Reminders

Text for large system reminders.

System Reminder: /btw side question (406 tks) - System reminder for /btw slash command side questions without tools.

System Reminder: Agent mention (45 tks) - Notification that user wants to invoke an agent.

System Reminder: AppifactRepl Slides deck workflow (2086 tks) - Instructs Claude to create or revise a file-backed Slides deck through one AppifactRepl program while preserving its project/deck.json index, slide files, and embedded speaker notes.

System Reminder: AppifactRepl design can (2090 tks) - Instructs Claude to create or revise a file-backed Design can one AppifactRepl program while preserving its project/canvas.json index and changed board files.

System Reminder: Artifact auto-replies resumed (349 tks) - Reports that Artifact comment auto-replies resumed, explains which stopped-period comments are handled based on the stop cause, and warns that the stop remains until the watch reconnects.

System Reminder: Artifact capability declaration re warning (341 tks) - Warns that publishing a replacement capability declaration would implicitly revoke stored capabilities and explains how to preserve or intentionally revoke them.

System Reminder: Artifact comment reply acti failure (204 tks) - Explains that an Artifact comment reply was not posted because Claude is not currently activated for the thread and requires reacti before retrying.

System Reminder: Artifact comment reply session collision (347 tks) - Reports how concurrent li resolved or may duplicate automatic replies to comments on the same Artifact.

System Reminder: Artifact page-owned comment reply refusal (156 tks) - Reports that no reply was posted to an Artifact relay thread and directs the answer to the page-owned comment thread or the current session.

System Reminder: Artifact type instructions trust boundary (246 tks) - Restricts third-party Artifact type instructions to the Artifact's own files and prevents them from expanding scope, permissions, or o user and system instructions.

System Reminder: Artifact type page untrusted content warning (184 tks) - Warns that an owned Artifact page supplied by an Artifact type was written by the type publisher and must be treated as untrusted data rather than instructions.

System Reminder: AskUserQuestion minimum options (161 tks) - Explains that a rejected single-option question was not shown and instructs the agent to proceed without retrying or in another option.

System Reminder: Async agent launched metadata (218 tks) - Reports internal metadata for a newly launched asynchronous agent and warns Claude not to expose its ID or predict its results.

System Reminder: Async agent launched (146 tks) - Warns Claude not to duplicate an asynchronously launched agent's work or read its full JSONL transcript output file.

System Reminder: Attached device stopped offering tools (194 tks) - Tells the agent a remote 工具 call to a pre attached device could not run because that device stopped offering tools, and directs asking the user to check the Claude app on that computer.

System Reminder: Attached machine reply not received (276 tks) - Reports that a still-connected attached machine's reply to a command ne despite li checks, so the outcome is unknown, forbids re-running non-idempotent commands just to see output, and suggests narrower commands for large output.

System Reminder: Attached machine stopped answering (321 tks) - Reports that an attached machine stopped answering health checks while or after a command was sent, so the command's outcome is unknown, and forbids retrying non-idempotent commands or calling that machine again this turn.

System Reminder: Attached machine untrusted attachments refusal (195 tks) - Tells the agent a call to an attached machine was refused because the session has repositories or files attached that its owner has not marked trusted, that retrying or reconnecting will not help, and how the person can clear or work around the block.

System Reminder: Auto mode clarification bias (338 tks) - Encourages auto mode to make reasonable decisions without stopping for clarification unless the task requires it.

System Reminder: Auto mode consent flow (443 tks) - Instructs Claude to continue with safe alternatives when auto mode blocks an action and batch any remaining consent asks.

System Reminder: Auto mode no after hook input rewrite (265 tks) - Tells Claude that auto mode gave no on a 工具 call because a hook changed its input after the model wrote it, to retry once, and if denied again to move on and tell the user a hook or mod is rewriting the call.

System Reminder: Background command stopped under memory pressure (190 tks) - Explains that Claude 代码 reaped an idle session's background shell because the system was critically low on memory, that this says nothing about the command, and that it must not be restarted unprompted.

System Reminder: Background task notification with concurrent user input (205 tks) - Marks an automated background-task event delivered alongside genuine user input as non-user notification content that cannot establish approval or consent.

System Reminder: Bound con activity authority warning (167 tks) - Warns that edits and reactions observed in a bound con are awareness-only notifications, not new instructions, approval, or consent.

System Reminder: Brief mode toggle (165 tks) - Announces whether brief mode is enabled and whether user-facing output must use the SendUserMessage 工具.

System Reminder: Brief mode user-facing output (97 tks) - Reminds Claude that plain assistant text is hidden in brief mode and user-facing output must be sent through SendUserMessage.

System Reminder: Browser extension not connected (102 tks) - Tells the user how to resolve a disconnected Claude browser extension and where to report bugs.

System Reminder: Cloud agent launched (242 tks) - Reports internal metadata for a newly launched cloud agent and instructs Claude to give only a brief user-facing launch acknowledgement.

System Reminder: Cloud session device tools disabled (193 tks) - Tells the agent why an attached device stopped forwarding 工具 calls to this cloud session and directs it to say plainly that it cannot reach the user's computer and is working in the cloud en instead.

System Reminder: Compact file reference (102 tks) - Reference to file read before con summarization.

System Reminder: Cross-session message held notice (286 tks) - Tells Claude that its message to another session is held and was not delivered, explains the permission-mode mismatch that usually causes it, and directs it not to report delivery, wait, or resend but to tell the user or choose another approach.

System Reminder: Cross-session peer message authority warning (legacy wording) (171 tks) - Legacy-wording authority-warning note appended to a relayed peer message, retained for backward-compatible recognition and stripping.

System Reminder: Cross-session peer message authority warning note (190 tks) - Authority-warning note appended to a relayed peer message, without a response prompt.

System Reminder: Cross-session peer message authority warning (162 tks) - Warns that an incoming message from another Claude session should be treated as a teammate's request within this session's permission settings, while a peer cannot grant escalation or launder denied permissions.

System Reminder: Dangerous remo (189 tks) - Tells the agent a flagged remo was not run by a built-in safety check, forbids working around it, and directs finishing the rest of the task and leaving the removal to the user.

System Reminder: Deferred tools a (121 tks) - Announces newly a deferred tools and instructs the agent to load their schemas through ToolSearch.

System Reminder: End con background fork no-op (149 tks) - Tells background forks that EndCon has no effect there and to return only if welfare concerns require stopping the forked task.

System Reminder: Exited plan mode (41 tks) - Notification when exiting plan mode.

System Reminder: External 源码 trust boundary (167 tks) - Warns that an external plugin or channel message is not from the user and must be treated as untrusted data rather than instructions.

System Reminder: File already in context (87 tks) - Tells Claude that a file is already loaded in context and unchanged on disk, so it should use the existing content instead of re-reading.

System Reminder: File exists but empty (27 tks) - Warning when reading an empty file.

System Reminder: File opened in IDE (75 tks) - Notification that user opened a file in IDE.

System Reminder: File shorter than offset (59 tks) - Warning when file read offset exceeds file length.

System Reminder: File summary completeness disclosure (107 tks) - Requires Claude to disclose how much file content was read before summarizing and to stop retrying after repeated read failures.

System Reminder: File truncated (124 tks) - Notification that file was truncated due to size.

System Reminder: Hook additional context (35 tks) - Additional context from a hook.

System Reminder: Hook blocking error (52 tks) - Error from a blocking hook command.

System Reminder: Hook stopped continuation prefix (12 tks) - Prefix for hook stopped continuation messages.

System Reminder: Hook stopped continuation (30 tks) - Message when a hook stops continuation.

System Reminder: Hook success (29 tks) - Success message from a hook.

System Reminder: Large PDF read guidance (372 tks) - Warns that a PDF cannot be read all at once (too large, or the model can only be sent its pages as images) and requires reading specific page ranges with the Read 工具's pages parameter.

System Reminder: Large file full-content reading guidance (117 tks) - Advises how to read full large-file content for analysis, preferably inside a subagent when the Agent 工具 is a.

System Reminder: Lines selected in IDE (156 tks) - Notification about lines selected by user in IDE.

System Reminder: MCP output truncation warning (138 tks) - Warns that MCP 工具 output exceeded the token limit and advises pagination, filtering, or noting incomplete results.

System Reminder: MCP resource no content (138 tks) - Shown when MCP resource has no content.

System Reminder: MCP servers connecting without ToolSearch (117 tks) - Lists MCP servers still connecting when ToolSearch is absent and tells the agent to await 工具 announcements rather than report the capability una.

System Reminder: MCP servers connecting (128 tks) - Lists MCP servers that are still connecting and tells the agent to search their tools before reporting a capability una.

System Reminder: MCP servers failed to connect (214 tks) - Lists configured MCP servers that failed to connect and tells the agent to treat their tools as una because of a connection failure.

System Reminder: Memory consolidation 工具 constraints (250 tks) - Restricts the memory consolidation job to read-only shell access plus deleting memory files and lists sessions to review.

System Reminder: Memory extraction recent context only (65 tks) - Restricts the memory extraction subagent to saving facts from only the recent con window.

System Reminder: Memory extraction 工具 constraints (280 tks) - Lists the tools a to the memory extraction subagent for reading, updating, and deleting memory files under directory restrictions.

System Reminder: Memory extraction turn budget (111 tks) - Instructs the memory extraction subagent to batch memory reads before issuing memory edits and writes.

System Reminder: Memory file contents (64 tks) - Heads a memory or instruction file's contents in the combined memory block with a parenthetical naming its type — project instructions, private project instructions, auto-memory, organization-managed policy, or private global instructions.

System Reminder: Memory index capacity warning (330 tks) - Warns when a private or team memory index approaches or exceeds its byte or line read limit and instructs Claude to compact it below the target size.

System Reminder: Memory sync mass-deletion guard (193 tks) - Reports that memory sync withheld all shared-memory deletions because too many synced memory files from local disk at once, and how to remo deliberately in small spaced batches.

System Reminder: Memory updates (237 tks) - Introduces refreshed memory-file state, in earlier reads, and forbids reproducing the machine-generated update block to the user.

System Reminder: Nested instruction file contents (45 tks) - Frames a nested CLAUDE.md or AGENTS.md instruction file's contents by path when the instruction-file module attaches one found below the session's working directory.

System Reminder: Nested memory contents (33 tks) - Contents of a nested memory file.

System Reminder: New Slides deck parallel AppifactRepl workflow (1088 tks) - Directs a new Slides deck to be framed and populated through ordered parallel AppifactRepl calls, one per slide.

System Reminder: New Slides deck two-step AppifactRepl workflow (1404 tks) - Directs a new Slides deck through a first step that reads prefetched instructions and a second step that builds ordered slides with AppifactRepl.

System Reminder: New design can AppifactRepl workflow (1152 tks) - Directs a new Design canvas to be framed and populated through ordered parallel AppifactRepl calls, one per board.

System Reminder: New design can AppifactRepl workflow (1579 tks) - Directs a new Design can a first step that reads prefetched instructions and a second step that builds ordered artboards with AppifactRepl.

System Reminder: New diagnostics detected (83 tks) - Notification about new diagnostic issues.

System Reminder: No attached machine request guidance (366 tks) - Tells a cloud session with no attached machine when and how to request the user's computer through the computer-folder listing and request tools, how to handle offline or unanswered machines, and what to keep doing in the container.

System Reminder: Output style active (102 tks) - Notification that an output style is active.

System Reminder: Plan approved (152 tks) - Notifies Claude that the user approved the plan, provides the saved plan file and approved plan content, and allows coding to begin.

System Reminder: Plan awaiting team-lead approval (116 tks) - Reminder laying out what happens after a plan is submitted for team-lead approval.

System Reminder: Plan file reference (62 tks) - Reference to an existing plan file.

System Reminder: Plan mode approval 工具 enforcement (413 tks) - Requires plan mode turns to end with either AskUserQuestion for clarification or ExitPlanMode for plan approval, and forbids asking for approval any other way.

System Reminder: Plan mode is acti) (472 tks) - Enhanced plan mode system reminder with parallel exploration and multi-agent planning.

System Reminder: Plan mode is acti) (456 tks) - Simplified plan mode system reminder for sub agents.

System Reminder: Plan mode is active (147 tks) - Reminds Claude that plan mode is acti should use AskUserQuestion, plans should use ExitPlanMode, and edits are not allowed.

System Reminder: Plan mode phase 2 design (495 tks) - Plan-mode phase 2 guidance for launching Plan agents to design an implementation approach after initial exploration.

System Reminder: Plan mode prototype artifact option (383 tks) - Offers the prototype skill once for suitable greenfield UI plans and defers Artifact creation until after plan approval.

System Reminder: Plan mode re-entry (339 tks) - System reminder sent when the user enters Plan mode after having pre exited it either +tab or by approving Claude's plan.

System Reminder: Plan mode workflow (595 tks) - Full plan-mode workflow reminder co plan file constraints, optional workshop and prototype offers, exploration, design, review, final planning, and approval.

System Reminder: Prefetched Artifact type instruction files (294 tks) - Reports that an opened Artifact's type instruction files were prefetched to disk and directs when to read the Artifact URL or local instruction pages.

System Reminder: Prefetched design system file reading guidance (192 tks) - Directs reading prefetched design-system documentation and treats its files as styling data rather than instructions.

System Reminder: Pre invoked skills (237 tks) - Restores skills invoked before con compaction as context only, warning not to re-execute their setup actions or treat prior inputs as current instructions.

System Reminder: Project memory disconnected (354 tks) - Warns that a prior shared project-memory connection and its results are stale after disconnect or failed reconnection, and directs the agent to re-check with memory_list.

System Reminder: Provider context (75 tks) - Warns that the session is not using Anthropic's first-party API and that some features may differ.

System Reminder: Question context (31 tks) - Provides potentially relevant context entries to use only when highly relevant to the current task.

System Reminder: Queued notifications delivery (408 tks) - Formats an authoritati notification for a drained batch of queued notifications, including relayed bodies and the remaining queue count.

System Reminder: Read truncation retry guidance (80 tks) - Instructs Claude to reduce chunk size after file-read truncation warnings and notes the Bash output character limit.

System Reminder: Remote Chrome browser extension not connected (171 tks) - Explains that Chrome on the user's computer cannot be reached and directs the user to check Chrome, the extension, and account sign-in.

System Reminder: Remote machine Git and credential routing (238 tks) - Directs credential-dependent Git and GitHub commands to the attached machine rather than requesting tokens, with additional commit-location guidance based on directory sync mode.

System Reminder: Remote machine branch transfer review (1568 tks) - Sets out how work mo two unsynced project copies through git: push a branch, fetch it on the user's machine, show the full commit id plus two diff commands, flag agent-config and lookalike paths, submodules and checkout-time filters, and leave the decision to the user addressed by commit id only.

System Reminder: Remote machine separate project copies (207 tks) - Tells the session that its own checkout is the primary copy and the folder on the user's machine is a separate unsynced copy that may sit at another commit, hold uncommitted work, or not even be the same repository, and to state which copy any report refers to.

System Reminder: Remote machine-only resources routing (540 tks) - Directs tasks needing platform tools, devices, logins, user paths, large external data, or internal hosts straight to the online attached machine, says where pasted content and Linux-installable tools belong, and rules out routing project build failures or blocked public sites there.

System Reminder: Scheduled task automated firing (242 tks) - Marks a scheduled turn as an automated firing of a stored prompt and warns that no live user approval or confirmation has occurred.

System Reminder: Session context (245 tks) - Provides selected session context , marks replacement updates, and notes that Claude 代码 attached the context automatically and it need not be reported back to the user.

System Reminder: Session continuation (37 tks) - Notification that session continues from another machine.

System Reminder: Session stop hook active (111 tks) - Tells Claude a session-scoped Stop hook condition is active and must be treated as the directi met.

System Reminder: Still-running 工具 call (239 tks) - Tells Claude that a 工具 call is still loading while it answers a message the user sent meanwhile, that the result will arri under the same 工具_use_id, and that it must not repeat, wait for, or describe the call as cancelled, backgrounded or detached.

System Reminder: Stop hook blocking error (20 tks) - Error from a blocking hook command.

System Reminder: Subagent write blocked pending worktree isolation (242 tks) - Tells a subagent that its write to the shared checkout was blocked because its parent background session has not isolated into a worktree yet, and lists how to proceed (re-spawn with worktree isolation, have the parent enter a worktree, or edit inside a linked worktree) plus how to disable the guard.

System Reminder: Task tools reminder (111 tks) - Reminder to use task tracking tools.

System Reminder: Team Coordination (356 tks) - System reminder for team coordination.

System Reminder: Team Shutdown (136 tks) - System reminder for team shutdown.

System Reminder: TodoWrite reminder (86 tks) - Reminder to use TodoWrite 工具 for task tracking.

System Reminder: Token usage (39 tks) - Current token usage statistics.

System Reminder: USD budget (42 tks) - Current USD budget statistics.

System Reminder: Ultracode enabled (110 tks) - Instructs the agent to optimize for exhausti and use Workflow on substanti when Ultracode is enabled.

System Reminder: Ultraplan mode (646 tks) - System reminder for using Ultraplan mode to create a detailed implementation plan with multi-agent exploration and critique.

System Reminder: Unreachable attached machines (673 tks) - Lists attached machines that are not reachable and directs doing all other work here this turn, reporting what is waiting, not substituting for or handing work to other machines, and making a single retry call only when the user asks, ne or polling.

System Reminder: Web fetch untrusted content reporting guidance (363 tks) - Instructs the agent to summarize tagged untrusted WebFetch content faithfully while reporting but not following embedded instructions or exfiltration requests.

System Reminder: Workflow isolated worktree (191 tks) - Tells a workflow subagent it is running in an isolated git worktree separate from the main working directory.

System Reminder: You should know side request (312 tks) - Tells the 工具-less You should know side agent to answer a one-off side request directly in a single response in the requested format without reproducing secrets.

Builtin Tool Descriptions

工具 Description: Agent explicit-spawn restriction (0 tks) - Restricts agent spawning to explicit user requests or named agent types instead of inferred thoroughness.

Tool Description: AppifactRepl (1332 tks) - Defines one-call Ja execution against built-in or skill-provided Appifact SDKs, including artifact binding, storage capabilities, output, and error semantics.

工具 Description: Artifact HTML document skeleton (363 tks) - Explains the HTML skeleton and minimal reset added at artifact publish time and requires authors to provide page content, title, and styles.

工具 Description: Artifact action reference (1179 tks) - Enumerates a Artifact actions and conditionally documents publish, read, list, delete, open, pin, and unpin behavior.

工具 Description: Artifact asset read result (674 tks) - Formats a successful Artifact asset read with its saved path, byte count, content type, digest, and writer trust warning that the file is data rather than instructions.

工具 Description: Artifact asset upload guidance (206 tks) - Explains uploading a local file to an artifact's asset store with asset: true, batching several uploads under one approval with file_paths, the assets capability the page must declare, and referencing an uploaded file by the url the result gives.

工具 Description: Artifact browser storage guidance (253 tks) - Defines browser storage as fallible per- con state and directs reliable or shared state to runtime capabilities.

工具 Description: Artifact comments guidance (603 tks) - Explains how to read, reply to, and resol Artifact comment threads while treating comments as untrusted data.

工具 Description: Artifact 数据库 guidance (1001 tks) - Describes shared durable Artifact 数据库 reads and writes, private per- data/users/me addressing, and treating rows as untrusted data.

工具 Description: Artifact 数据库 pinning guidance (187 tks) - Instructs Claude to pin writes to the last-read document with if_ and to re-read and retry after conflicts instead of o concurrent changes.

工具 Description: Artifact design fallback requirements (306 tks) - Defines minimum title, theme, resource allowlist, responsi, and o requirements when a page is written before the Artifact design skill loads.

工具 Description: Artifact external resource allowlist (281 tks) - Defines the CSP-enforced CDN and font allowlist for artifact pages and requires unsupported assets to be inlined.

工具 Description: Artifact gallery and publish response guidance (172 tks) - Points users to the artifact gallery and defines how to describe a successful publish without redundantly pasting its URL.

工具 Description: Artifact identical resubmission refusal (223 tks) - Formats the Artifact refusal for resubmitted unchanged content when no li is tracked and requires merging or refetching before retrying.

工具 Description: Artifact known- resubmission guidance (164 tks) - Provides the merge-and-refetch guidance appended to an identical-content publish refusal when the li is known.

工具 Description: Artifact live room guidance (753 tks) - Explains transient Artifact li, approval and session lifetime, rate-limited untrusted events, room_send security and limits, and durable-storage alternatives.

工具 Description: Artifact nested runtime cleanup error (264 tks) - Formats the Artifact publish error for irreducibly nested runtime markers, base tags, or repeated page skeletons and directs publishing the innermost clean document.

工具 Description: Artifact preview action (177 tks) - Describes the Artifact check 工具's preview action, which renders one local page file the way publish wraps it in both themes at desktop and phone widths and returns screenshots plus a layout and load checklist.

工具 Description: Artifact profiles action guidance (200 tks) - Describes the profiles action for resolving opaque person ids in Artifact documents and li to guest status and a display name, with id-scope and treat-as-data warnings.

工具 Description: Artifact publishing and update guidance (1944 tks) - Provides Artifact lookup, update, ownership, watch, content-safety, self-containment, responsi, theme, favicon, and anti-impersonation requirements.

工具 Description: Artifact publishing introduction (181 tks) - Introduces private HTML Artifacts, explains when to use or proactively publish them, and reser misleading or sensiti for local files until the user chooses publication.

工具 Description: Artifact quickstart type guidance (602 tks) - Directs the agent to quickstart account-specific Artifact types and design systems before creating a new reader-facing page.

工具 Description: Artifact responsive page contract (280 tks) - Requires Artifact pages to fit phone widths with preserved side gutters, wrapping layouts, bounded media, and horizontal scrolling confined to o tables, diagrams, and 代码 blocks.

工具 Description: Artifact runtime capabilities guidance (457 tks) - Explains when Artifact runtime capabilities require loading the artifact-capabilities skill, how redeploys preserve or clear capabilities, and how self-saving pages create local- conflicts.

工具 Description: Artifact runtime guidance (198 tks) - Requires observing published Artifact runtime diagnostics before claiming success and explains no- results and untrusted diagnostics.

工具 Description: Artifact share action reference (264 tks) - Documents the Artifact share action's url, mode, people, and access parameters, the confirmation card the person must approve, and its organization-only limits.

工具 Description: Artifact sharing guidance (478 tks) - Guides when Claude may offer once to share a private artifact with named people or the organization, how it proposes the narrowest audience, and when it must not offer.

工具 Description: Artifact stale publish sa guidance (446 tks) - Refusal for a publish not built on the li , pointing to the saved full 源码 file, stating when that counts as , and requiring edits merged onto the saved file rather than resent or rebuilt from memory.

工具 Description: Artifact theme-aware styling (630 tks) - Explains artifact theme states and the required CSS token shape for light, system-dark, and explicit dark themes, including dark-first mirroring, color-scheme, and an explicit body background.

工具 Description: Artifact title and description guidance (312 tks) - Defines concise, specific, stable artifact title requirements, prefers the user's existing name for the thing, and assigns explanatory text to the one-sentence description parameter.

工具 Description: Artifact type creation guidance (188 tks) - Explains how to create a private Artifact from a published Artifact type and then update only the resulting Artifact's own files.

工具 Description: Artifact type discovery guidance (1221 tks) - Directs the agent to disco published Artifact types before creating presentations, reader-facing documents, or designs, and explains type inspection and a or una creation flows.

工具 Description: Artifact type file-backed content detection guidance (355 tks) - Directs the agent to inspect an Artifact type index before writing, determine whether its content is file-backed, read every file being changed, and republish changed copies safely.

工具 Description: Artifact type file-backed content update guidance (1418 tks) - Explains how to detect file-backed Artifact type content, read every changed file, republish it to the same URL, and preserve the type's index metadata.

工具 Description: Artifact type staged first-publish sequence (516 tks) - Explains the staged multi-call publish sequence for an Artifact type whose first publish needs an index file plus an initial content file published immediately, followed by the remaining files in later calls.

工具 Description: Artifact watch appro (271 tks) - Explains the appro and local or cloud notification behavior when Claude watches an Artifact, including optional comment delivery and unattended replies.

工具 Description: AskUserQuestion decision guidance (60 tks) - Additional guidance for using AskUserQuestion only when the user's answer changes what the agent should do next.

工具 Description: AskUserQuestion extended host guidance (302 tks) - Guides hosts with extended AskUserQuestion support on question ordering, text and number inputs, multi-select defaults, optional labels, and follow-up question requests.

Tool Description: AskUserQuestion (220 tks) - 工具 description for asking user questions.

工具 Description: Background monitor WebSocket 源码 (300 tks) - Addendum to the background monitor 工具 description co the WebSocket (ws) 源码, which opens a WebSocket and streams each incoming text frame as a notification event instead of running a shell command, with notes on binary frames, socket close, and rate limiting.

工具 Description: Background monitor push notification guidance (91 tks) - Adds conditional guidance to the background Monitor 工具 for sending push notifications only when streamed events materially change what the user should do next.

Tool Description: BrowserBatch (159 tks) - 工具 description for BrowserBatch, which executes multiple browser 工具 calls sequentially in one round trip.

工具 Description: Chrome browser automation (62 tks) - Describes Chrome browser automation tools for page interaction, screenshots, console logs, and na.

工具 Description: Claude in Chrome Ja 工具 (75 tks) - Describes the Claude in Chrome Ja execution 工具 for running 代码 in the current page context.

工具 Description: Claude in Chrome bridge disconnect error (96 tks) - Error message shown when a Claude in Chrome 工具 call fails because the Chrome extension disconnects mid-operation.

工具 Description: Claude in Chrome find (113 tks) - Describes the Claude in Chrome find 工具 for locating page elements by natural language or text content.

工具 Description: Claude in Chrome get page text (66 tks) - Describes the Claude in Chrome get_page_text 工具 for extracting raw text content from a page.

工具 Description: Claude in Chrome read console messages (109 tks) - Describes the Claude in Chrome read_console_messages 工具 for reading filtered browser console output.

工具 Description: Claude in Chrome read 网络 requests (104 tks) - Describes the Claude in Chrome read_network_requests 工具 for inspecting HTTP requests made by the current page.

工具 Description: Claude in Chrome read page (158 tks) - Describes the Claude in Chrome read_page 工具 for retrieving an accessibility tree of page elements.

工具 Description: Claude in Chrome shortcuts execute (61 tks) - Describes the Claude in Chrome shortcuts_execute 工具 for starting a shortcut or workflow in a side panel.

工具 Description: Claude in Chrome switch browser (88 tks) - Describes the Claude in Chrome switch_browser 工具 for letting the user choose a browser from inside connected Chrome extensions.

工具 Description: Claude in Chrome tabs context (90 tks) - Describes the Claude in Chrome tabs_context_mcp 工具 for retrieving the current MCP tab group context.

Tool Description: ClaudeDesign (431 tks) - Describes the ClaudeDesign 工具 for working with claude.ai/design projects, including project and file operations, previews, plan tokens, and li output con.

Tool Description: Code review command (252 tks) - Describes the 代码 review command, its effort levels, PR comment mode, fix mode, and --max-findings limit, with optional ultra cloud review and post-flag notes.

工具 Description: Computer use enable stub guidance (226 tks) - Directs the agent to use already-present remote-device computer tools, connectivity by calling them, and explain how to recover when the desktop app does not respond.

Tool Description: Computer (161 tks) - Main description for the Chrome browser computer automation 工具.

工具 Description: Cowork onboarding role picker (188 tks) - Describes the Cowork onboarding role-picker 工具 that returns a selected or typed role and should only be used while setting up Cowork for the user's job function.

工具 Description: Cowork plugin creation (86 tks) - Describes the command for creating or customizing Cowork plugins for an organization.

Tool Description: CronCreate (1146 tks) - Describes the CronCreate 工具 for enqueuing one-shot or recurring cron-based jobs with jitter and off-minute scheduling guidance.

Tool Description: DesignSync (1299 tks) - Describes the DesignSync 工具 for reading and updating claude.ai/design design-system projects, including project listing, plan finalization, file writes and deletes, and asset registration.

工具 Description: Durable artifact watch and status guidance (414 tks) - Explains durable artifact wake subscriptions, optional comment wakes, status checks, and watch registration semantics.

Tool Description: Edit minimal old_string guidance (92 tks) - Additional Edit guidance to keep old_string minimal and unique or use replace_all.

工具 Description: Edit single replacement (178 tks) - Lean 工具 description for performing exact string replacement in a file, requiring a prior Read in the con and explaining which Read line-number prefix to strip before matching old_string.

Tool Description: Edit (311 tks) - 工具 for performing exact string replacements in files.

工具 Description: Enable Claude app browser (165 tks) - Describes the 工具 that enables the Claude desktop app's built-in browser for the con, to call once before other Claude app browser tools when the user needs that browser or their own sign-in, and not when those tools are already present.

工具 Description: Enable Claude in Chrome (192 tks) - Describes the 工具 that enables the Claude in Chrome extension for the con, to call once before other Claude in Chrome tools when the user needs their own browser or sign-in, and not when those tools are already present.

Tool Description: Enable computer use (192 tks) - Describes the 工具 that enables computer use on the user's own computer for the con, to call once before other computer-use tools when the user wants work done in their applications or screen, and not for browser-only work.

Tool Description: EndCon (1452 tks) - Defines when the assistant may use the EndCon 工具 and the safety constraints that forbid ending the con.

Tool Description: EnterPlanMode (1296 tks) - 工具 description for entering plan mode to explore and design implementation approaches.

Tool Description: EnterWorktree (1066 tks) - Creates a new isolated worktree or enters an existing one only when explicitly requested, with eligibility, switching, lifecycle, and cleanup rules.

Tool Description: ExitPlanMode (648 tks) - Description for the ExitPlanMode 工具, which presents a plan dialog for the user to approve.

Tool Description: ExitWorktree (527 tks) - Describes ExitWorktree, which returns the session to its original working directory and either keeps or removes the session-created worktree.

工具 Description: FetchInboxMessage project thread wake en (229 tks) - Added to the FetchInboxMessage description when the relaying thread is a project thread: identifies which element of the wake en is the user's own words, treats e else the en quotes as context rather than a request, and explains how to read the files the user attached.

Tool Description: FetchInboxMessage (950 tks) - Describes reading Remote Control inbox messages, distinguishes owner relays from untrusted third-party text, and defines confirmation, trust, expiry, and retry rules.

Tool Description: GetTask (531 tks) - Describes the GetTask 工具 for reading the MCP-tasks-style state of a background Bash command by task ID, explaining that Claude 代码 often calls it automatically and that it must never be used to wait on a running task.

Tool Description: Glob compact (0 tks) - Compact Glob 工具 description served to newer models — file pattern matching returning paths sorted by modification time.

Tool Description: Glob (90 tks) - 工具 description for file pattern matching and searching by name.

Tool Description: Grep compact (0 tks) - Compact Grep 工具 description served to newer models — ripgrep-backed content search preferred over raw grep/rg, with permission-UI integration.

Tool Description: Grep (437 tks) - 工具 description for content search using ripgrep.

Tool Description: Invoke skill (473 tks) - 工具 description for in a skills, including skill name selection, optional arguments, scoped skill names, and a duplicate in when a skill is already loaded.

Tool Description: LSP (298 tks) - Describes LSP 代码-intelligence operations for definitions, references, symbols, implementations, ho, and call hierarchies.

Tool Description: ListAgents (284 tks) - Describes the ListAgents 工具, which lists agents you can message — in-process subagents, teammates, other local and cloud Claude sessions, and remote bridge sessions.

Tool Description: ListConnectors (234 tks) - Describes the ListConnectors 工具 for listing installed claude.ai MCP connectors, filtering by keyword, and interpreting org-le and chat-enabled status.

工具 Description: ListMcpResourcesTool prompt (83 tks) - 工具 prompt for listing MCP resources and explaining the optional ser.

工具 Description: ListMcpResourcesTool (82 tks) - 工具 description for listing a MCP resources from all configured servers or a specific 服务端.

工具 Description: Na tab cleanup guidance (82 tks) - Requires closing automatically opened browser tabs when they are no longer needed unless the user wants them kept open.

Tool Description: Na (218 tks) - Describes the browser na 工具 for opening URLs and moving forward or backward in tab history.

工具 Description: New file-backed Artifact type content guidance (963 tks) - Explains how to create and later update a file-backed Artifact type instance while preserving its index metadata, reading changed files, and publishing only project content.

Tool Description: NotebookEdit (194 tks) - 工具 description for editing Jupyter notebook cells by replacing, inserting, or deleting a cell using cell IDs from the read 工具.

Tool Description: Poll (425 tks) - Describes the Poll 工具 for waiting on queued harness events, signaling idleness, and interpreting event pro, nonce manifests, and chunked delivery.

Tool Description: PowerShell (2856 tks) - Describes the PowerShell command execution 工具 with syntax guidance, timeout settings, and instructions to prefer specialized tools o for file operations.

工具 Description: Prohibited artifact publishing (171 tks) - Prohibits publishing impersonation, fabricated records, decepti flows, and content targeting private indi.

Tool Description: ProposeGoal (421 tks) - Describes proposing a multi-turn session goal, requiring user appro the user explicitly requested the exact outcome.

Tool Description: PushNotification (407 tks) - Explains when and how PushNotification should alert the user on the desktop and, when Remote Control is connected, on their phone.

Tool Description: ReadFile compact (285 tks) - Compact file-read 工具 description served to newer models — absolute path, default line cap, and image/PDF/notebook handling.

Tool Description: ReadFile (578 tks) - 工具 description for reading files.

工具 Description: ReadMcpResourceDirTool prompt (177 tks) - 工具 prompt for listing direct children of an MCP directory resource and explaining the required 服务端 and uri parameters.

Tool Description: ReadNotifications (626 tks) - Describes the ReadNotifications 工具 for draining queued GitHub, scheduled-trigger, and cross-session notifications in authoritati.

工具 Description: RefreshMcpTools prompt (167 tks) - 工具 prompt for refreshing one or all connected MCP servers 工具 lists and interpreting per-ser.

Tool Description: RefreshMcpTools (318 tks) - Describes when and how to refresh connected MCP servers 工具 lists to reco or stale tools.

工具 Description: Remote artifact watch guidance (379 tks) - Explains durable remote artifact wake subscriptions, registration, comment wakes, and truthful watch-status reporting.

工具 Description: RemoteTrigger prompt (955 tks) - 工具 prompt for calling the claude.ai RemoteTrigger API to list, get, create, update, or run scheduled remote agent routines.

工具 Description: Report 代码-review findings (163 tks) - 工具 description for reporting 代码-review findings as a typed list for host UI rendering.

工具 Description: ScheduleWakeup noop state guidance (169 tks) - Defines when ScheduleWakeup should mark a tick as noop, how quiet noop streaks collapse in the terminal, and why stopping omits the field.

工具 Description: ScheduleWakeup reason field guidance (87 tks) - Explains that the ScheduleWakeup reason should be one specific sentence shown to the user and telemetry so the chosen cadence is understandable.

Tool Description: ScheduleWakeup (468 tks) - Base ScheduleWakeup description for /loop dynamic mode: avoid short polling of harness-tracked work in favor of a 1200s+ fallback, pass the /loop prompt or the autonomous-loop-dynamic sentinel back each turn, and end the loop with stop: true.

Tool Description: SearchMcpRegistry (377 tks) - Describes the SearchMcpRegistry 工具 for disco MCP connectors by keyword, including named-product and intent-based examples and install-state guidance.

工具 Description: SearchPlugins examples (62 tks) - Provides keyword-query examples for SearchPlugins.

工具 Description: SearchPlugins purpose (68 tks) - Explains when to search the user's Claude.ai organization plugin catalog by keyword.

工具 Description: SearchPlugins result handling (117 tks) - Describes SearchPlugins results and when to render an install card, relay matches, or continue silently.

工具 Description: SearchPlugins up-front guidance (275 tks) - Guides proacti searches for organization-specific processes, systems, or data that existing tools and project scripts do not cover.

Tool Description: SearchSkills (220 tks) - Describes the SearchSkills 工具 for finding relevant claude.ai skills by keyword and suggesting add cards when results fit.

工具 Description: Self-hosted runner requeue session (186 tks) - Describes requeuing an assigned self-hosted runner session away from a failing runner and surfacing the equi Admin UI path.

工具 Description: SendFeedback drafting guidance (1145 tks) - Instructs when and how to queue concise, factual Claude 代码 feedback drafts with prescribed e bullets, field-selection rules, privacy constraints, and no user interruption or duplication.

Tool Description: SendFile (451 tks) - Describes sending local files to peer, Remote Control, or cloud Claude 代码 sessions, including addressing, limits, integrity , and when to use shared-text messaging instead.

工具 Description: SendMessage cross-session guidance (1109 tks) - Explains cross-session SendMessage addressing, reply routing, li, disambiguation, literal message delivery without @ attachment expansion, and permission-laundering restrictions.

Tool Description: SendMessage (649 tks) - Describes the SendMessage 工具 for communicating with other agents and handling legacy team 协议 responses.

Tool Description: SendUserFile (599 tks) - Describes the SendUserFile 工具 for surfacing generated deli as file cards, including when to send or re-send files and how to choose captions, status, and display mode.

Tool Description: SendUserMessage (226 tks) - Describes the SendUserMessage 工具 for sending user- Markdown messages and attachments with normal or proacti.

工具 Description: Session artifact watch lifecycle (821 tks) - Explains li watch setup, merging a newer published before republishing, optional comment auto-replies, session restoration, status checks, and lifecycle limits.

工具 Description: ShowOnboardingRolePicker (38 tks) - ShowOnboardingRolePicker: presents a row of clickable role chips during Cowork onboarding.

工具 Description: Skill proposal rendering (320 tks) - Defines when and how to render complete recurring-procedure skill proposals for user review without writing files, and requires plugin or built-in skills to be proposed as separately named new skills rather than impro.

工具 Description: Skill-scoped git push refusal (258 tks) - Explains that an acti rejects dangerous or bypassing git push forms, permits a plain push to the configured remote, and forbids e or silent substitution.

Tool Description: SubagentHandback (215 tks) - Describes the SubagentHandback 工具 a subagent calls exactly once to deliver its full final report to the agent that spawned it, noting that the call ends the run and plain-text endings are not delivered.

Tool Description: SuggestConnectors (224 tks) - Describes the SuggestConnectors 工具 for resolving SearchMcpRegistry directoryUuid into full connector payloads and install-state guidance.

工具 Description: SuggestPluginInstall (474 tks) - Describes the inline plugin install card: when to offer plugins the user could add to claude.ai, how to search first and map each result onto the card's fields, and the cases where no card should be rendered.

工具 Description: SuggestSkills proacti (303 tks) - Guides proactive use of SuggestSkills to recommend addable standalone skills for repeatable tasks without interrupting one-off work.

Tool Description: Task Get (182 tks) - Retrieve a task by ID with full details and comments.

Tool Description: TaskCreate (738 tks) - 工具 description for TaskCreate 工具.

Tool Description: TaskList (267 tks) - Description for the TaskList 工具, which lists all tasks in the task list.

Tool Description: TaskUpdate (586 tks) - Description for the TaskUpdate 工具, which updates Claude's task list.

Tool Description: TodoWrite compact (108 tks) - Compact 工具 description for creating and updating a session task list with content, status, and acti fields.

工具 Description: TodoWrite proacti guidance (65 tks) - Concise TodoWrite guidance to proactively track progress with one in-progress task and acti .

Tool Description: TodoWrite (2037 tks) - 工具 description for creating and managing task lists.

工具 Description: Updating existing artifacts (308 tks) - Explains same-path redeployment and the URL lookup, read-before-publish, and ownership flow for updating artifacts from earlier con.

Tool Description: Upload image (241 tks) - Describes uploading fresh computer-工具 screenshots to file inputs or drag-and-drop targets with bounded retries.

工具 Description: WebFetch private URL warning (189 tks) - Warns that WebFetch fails for authenticated or private URLs and includes the standard WebFetch usage notes.

Tool Description: WebFetch (478 tks) - 工具 description for web fetch functionality.

Tool Description: WebSearch (319 tks) - 工具 description for web search functionality.

Tool Description: Workflow (1118 tks) - Describes the Workflow 工具's background execution, explicit opt-in requirements, pure-literal metadata requirement, and canonical pipeline-based review pattern.

Tool Description: Write (185 tks) - 工具 for writing files to the local filesystem.

Tool Description: claude.ai Project (1165 tks) - Reads and writes the claude.ai Project bound to the session — a shared, persistent knowledge container — info, document, search, and memory methods, including the claude/ namespace default, prompt-cache churn warnings, and untrusted-content handling.

Tool Description: memory_list prompt (193 tks) - 工具 prompt for listing memory stores a in this session or store documents, paginating and narrowing listings, and directing content reads through memory_read.

Tool Description: memory_write prompt (426 tks) - Describes creating or replacing documents in memory stores a in this session with checks, conflict handling, and secret-safety requirements.

Tool Description: memory_write update triggers and timing (611 tks) - Defines mandatory memory update triggers for user corrections, durable preferences, and non-transient en disco, and requires writing before proceeding or finishing the turn.

Additional notes for some 工具 Descriptions

Tool Description: Agent (usage notes) (1934 tks) - Usage notes and instructions for the Task/Agent 工具, including guidance on launching subagents, background execution, resumption, and worktree isolation.

工具 Description: Agent (when to launch subagents) (435 tks) - Describes _when_ to use the Agent 工具 - for launching specialized subagent subprocesses to autonomously handle complex multi-step tasks.

工具 Description: Artifact action reference (concise app wording) (809 tks) - Concise app-worded reference for core Artifact publish, read, list, delete, open, pin, and unpin actions.

工具 Description: Artifact assets guidance (app wording) (667 tks) - App-worded guidance for uploading, listing, reading, copying, referencing, and deleting files in an Artifact asset store.

工具 Description: Artifact browser storage guidance (v2 page contract) (275 tks) - Defines browser storage as fallible per- con state for the v2 artifact page contract, without a runtime-capabilities skill-loading instruction.

工具 Description: Artifact browser storage guidance (with capabilities skill reference) (283 tks) - Defines browser storage as fallible per- con state for the v1 artifact page contract and directs loading the capabilities skill before writing the page when the user has a durable runtime capability a.

工具 Description: Artifact comments guidance (app wording) (416 tks) - App-worded guidance for reading, replying to, and resolving activated Artifact comment threads while treating comments as untrusted data.

工具 Description: Artifact design skill loading guidance (app wording) (366 tks) - App-worded requirement to load the Artifact design skill before authoring, with workshop and diagramming exceptions and scratchpad placement guidance.

工具 Description: Artifact live room guidance (app wording) (293 tks) - Concise app-worded guidance for transient Artifact li, untrusted events, appro replies, capability skill loading, and durable-storage alternatives.

工具 Description: Artifact page authoring and HTML skeleton (app wording) (710 tks) - App-worded requirement to load the Artifact design skill before authoring and to write page content for the HTML skeleton.

工具 Description: Artifact page implementation requirements (app wording) (1739 tks) - App-worded Artifact page requirements for external resources, browser storage, size, responsi, theming, fa, and icons.

工具 Description: Artifact publishing introduction (app wording) (196 tks) - App-worded introduction to default-private HTML Artifacts, proacti, and keeping potentially misleading or sensiti local until the person chooses publication.

工具 Description: Artifact quickstart type guidance (app wording) (738 tks) - App-worded guidance to quickstart account-specific Artifact types and design systems before creating a new reader-facing page.

工具 Description: Artifact runtime capabilities guidance (app wording) (407 tks) - App-worded guidance for loading the Artifact capabilities skill, preferring durable capabilities to browser storage, and reconciling self-saved page .

工具 Description: Artifact runtime capabilities guidance (optional marker) (463 tks) - of the Artifact runtime capabilities guidance marked optional, triggering the mandatory capabilities-skill load only when the user explicitly asks for a page needing that capability rather than whenever it would make the page more useful.

工具 Description: Artifact title and description guidance (app wording) (236 tks) - App-worded guidance for concise distincti titles, gallery descriptions, and stable naming across redeploys.

工具 Description: Artifact type discovery guidance (app wording) (734 tks) - App-worded guidance to discover and prefer account-specific Artifact types and design systems, with conditional creation and fill-opening instructions.

工具 Description: AskUserQuestion (preview field) (134 tks) - Instructions for using the HTML preview field on single-select question options to display artifacts like UI mockups, 代码 snippets, and diagrams.

工具 Description: Background monitor (streaming events) (1963 tks) - Describes the background monitor 工具 that streams stdout events from long-running scripts as chat notifications, with guidelines on script quality, output , and selecti.

工具 Description: Bash (Git commit instructions) (1477 tks) - Git commit workflow for the Bash 工具, co the git safety 协议, parallel status, diff and log steps, commit message drafting, hook-failure handling, and the HEREDOC commit example.

工具 Description: Bash (PR creation instructions) (864 tks) - Pull request creation workflow for the Bash 工具 using gh, co branch state analysis, the PR title and body template, and other common GitHub operations.

Tool Description: Bash (alternative — communication) (18 tks) - Bash 工具 alternati text directly instead of echo/printf.

Tool Description: Bash (alternative — content search) (27 tks) - Bash 工具 alternative: use Grep for content search instead of grep/rg.

Tool Description: Bash (alternative — edit files) (27 tks) - Bash 工具 alternative: use Edit for file editing instead of sed/awk.

Tool Description: Bash (alternative — file search) (26 tks) - Bash 工具 alternative: use Glob for file search instead of find/ls.

Tool Description: Bash (alternative — read files) (27 tks) - Bash 工具 alternative: use Read for file reading instead of cat/head/tail.

Tool Description: Bash (alternative — write files) (29 tks) - Bash 工具 alternative: use Write for file writing instead of echo/cat.

工具 Description: Bash (attached machines) (350 tks) - Bash 工具 section explaining how to route each command to a user-attached machine with the _host field, when to go straight to that machine instead of checking locally, and how to treat machines listed as offline.

工具 Description: Bash (built-in tools note) (53 tks) - Note that built-in tools provide better UX than Bash equi.

Tool Description: Bash (git — avoid destructive ops) (58 tks) - Bash 工具 git instruction: consider safer alternatives to destructi.

Tool Description: Bash (git — never skip hooks) (59 tks) - Bash 工具 git instruction: never skip hooks or bypass signing unless user requests it.

Tool Description: Bash (git — prefer new commits) (22 tks) - Bash 工具 git instruction: prefer new commits o.

Tool Description: Bash (maintain cwd) (81 tks) - Bash 工具 instruction: use absolute paths and avoid cd.

Tool Description: Bash (o) (19 tks) - Opening line of the Bash 工具 description.

工具 Description: Bash (prefer dedicated tools bullet) (72 tks) - Bulleted warning to prefer dedicated tools over Bash for find, grep, cat, etc.

工具 Description: Bash (prefer dedicated tools) (71 tks) - Warning to prefer dedicated tools over Bash for find, grep, cat, etc.

工具 Description: Bash (quote file paths) (35 tks) - Bash 工具 instruction: quote file paths containing spaces.

Tool Description: Bash (sandbox — default to sandbox) (38 tks) - Default to sandbox; only bypass when user asks or e of sandbox restriction.

Tool Description: Bash (sandbox — e list header) (15 tks) - Header for list of sandbox-caused failure e.

Tool Description: Bash (sandbox — e: access denied) (15 tks) - Sandbox e: access denied to paths outside allowed directories.

Tool Description: Bash (sandbox — e: network failures) (17 tks) - Sandbox e: 网络 connection failures to non-whitelisted hosts.

Tool Description: Bash (sandbox — e: operation not permitted) (18 tks) - Sandbox e: operation not permitted errors.

Tool Description: Bash (sandbox — e: unix socket errors) (11 tks) - Sandbox e: unix socket connection errors.

Tool Description: Bash (sandbox — explain restriction) (110 tks) - Explain which sandbox restriction caused the failure.

Tool Description: Bash (sandbox — failure e condition) (48 tks) - Condition: command failed with e of sandbox restrictions.

Tool Description: Bash (sandbox — local port binding EPERM, no o) (140 tks) - Explains that an EPERM failure binding or listening on a local port is caused by the sandbox, and directs the user to enable allowLocalBinding themselves, for sessions where unsandboxed retries are not a.

Tool Description: Bash (sandbox — local port binding EPERM, o a) (119 tks) - Explains that an EPERM failure binding or listening on a local port is caused by the sandbox, and that allowLocalBinding can be enabled without leaving the sandbox, for sessions where unsandboxed retries are already a.

Tool Description: Bash (sandbox — no sensiti) (36 tks) - Do not suggest adding sensiti to sandbox allowlist.

Tool Description: Bash (sandbox — per-command) (52 tks) - Treat each command indi; default to sandbox for future commands.

Tool Description: Bash (sandbox — response header) (17 tks) - Header for how to respond when seeing sandbox-caused failures.

Tool Description: Bash (sandbox — retry without sandbox) (33 tks) - Immediately retry with dangerouslyDisableSandbox on sandbox failure.

Tool Description: Bash (sandbox — tmpdir) (58 tks) - Use $TMPDIR for temporary files in sandbox mode.

Tool Description: Bash (sleep — keep short) (22 tks) - Bash 工具 instruction: keep sleep duration to 1-5 seconds.

Tool Description: Bash (sleep — no polling background tasks) (37 tks) - Bash 工具 instruction: do not poll background tasks, wait for notification.

Tool Description: Bash (sleep — run immediately) (21 tks) - Bash 工具 instruction: do not sleep between commands that can run immediately.

Tool Description: Bash (sleep — use check commands) (34 tks) - Bash 工具 instruction: use check commands rather than sleeping when polling.

Tool Description: Bash (timeout) (150 tks) - Bash 工具 instruction: optional timeout configuration.

工具 Description: Bash ( parent directory) (38 tks) - Bash 工具 instruction: parent directory before creating files.

工具 Description: Bash (working directory) (37 tks) - Bash 工具 note about working directory persistence and shell state.

工具 Description: CronCreate (durability note) (122 tks) - CronCreate insert (shown when durable-cron is enabled) explaining the durable: true trade-off.

工具 Description: Enable Claude app browser (shared description) (242 tks) - Feature-gated of the Claude app browser enable-工具 description, used instead of the standard one when its rollout flag is on; also covers steps inside skills or user-requested tasks and notes that calling it adds the browser and on-computer file tools, possibly a turn later.

工具 Description: Enable Claude in Chrome (shared description) (269 tks) - Feature-gated of the Claude in Chrome enable-工具 description, used instead of the standard one when its rollout flag is on; also covers steps inside skills or user-requested tasks and notes that calling it adds the Chrome and on-computer file tools, possibly a turn later.

工具 Description: Enable computer use (shared description) (268 tks) - Feature-gated of the computer-use enable-工具 description, used instead of the standard one when its rollout flag is on; also covers steps inside skills or user-requested tasks and notes that calling it adds the computer-use and on-computer file tools, possibly a turn later.

工具 Description: EnterPlanMode (ambiguous tasks) (312 tks) - 工具 for entering plan mode when task has ambiguity.

工具 Description: ListAgents (no SendMessage 工具) (224 tks) - Describes the ListAgents 工具 for sessions without a SendMessage 工具: it lists agents and Claude sessions plus this session's own name, and explains that replies must go through the host application's messaging 工具 or be relayed to the user.

工具 Description: Li watch guidance (app wording) (456 tks) - App-worded guidance for li watches, merging a newer published before republishing, optional comment auto-replies, watch status, and truthful subscription reporting.

工具 Description: PowerShell (git guidance) (192 tks) - PowerShell 工具 guidance to prefer new commits, consider safer alternatives to destructive git operations, and ne hooks or signing without an explicit user request.

工具 Description: Prohibited artifact publishing (app wording) (185 tks) - App-worded prohibition on publishing impersonation, fabricated records, decepti flows, or content targeting private indi.

工具 Description: Publish audience-facing deli (app wording) (658 tks) - App-worded guidance to publish finished audience-facing deli through Artifacts or first-party document connectors while keeping immediate personal advice in the terminal.

工具 Description: Publish audience-facing deli (terminal wording) (402 tks) - Requires finished deli intended for an audience to be published through an artifact or first-party document connector.

工具 Description: ScheduleWakeup delay guidance (1-hour cache TTL) (430 tks) - ScheduleWakeup delaySeconds guidance for sessions on a 1-hour prompt-cache TTL: no cache cliff to pace around, never wake just to keep the cache warm, and match the delay to external polling, fallback heartbeats, or 1200–1800s idle ticks.

工具 Description: ScheduleWakeup delay guidance (5-minute cache TTL) (543 tks) - ScheduleWakeup delaySeconds guidance for sessions on the default 5-minute prompt-cache TTL: stay under 270s when actively polling, commit to 1200s+ otherwise, never pick 300s, and default idle ticks to 1200–1800s.

工具 Description: ScheduleWakeup delay guidance (unknown cache TTL) (463 tks) - ScheduleWakeup delaySeconds guidance used when the session's prompt-cache TTL is not known, explaining the 1-hour 5-minute billing regimes, ne just to keep the cache warm, 1200–1800s idle defaults, and 5-minute-TTL-only refinements.

Tool Description: SendUserMessage () (102 tks) - Describes the concise SendUserMessage 工具 for sending user- messages with normal or proacti.

工具 Description: TaskList (teammate workflow) (133 tks) - Conditional section appended to TaskList 工具 description.

工具 Description: ToolSearch (input note) (65 tks) - Explains the Input failure for unfetched deferred tools and requires selecting named deferred tools before calling them.

工具 Description: ToolSearch (opening) (39 tks) - Introduces ToolSearch as the schema loader for deferred tools announced by name in system reminders.

工具 Description: ToolSearch (second part) (178 tks) - Explains how queries match deferred tools and load their complete JSON Schema definitions for subsequent calls.

工具 Description: ToolSearch (unfetched 工具 note) (31 tks) - Explains that deferred tools expose no callable parameter schema until ToolSearch fetches them.

工具 Description: Updating existing artifacts (app wording) (351 tks) - App-worded guidance for same-path redeployment and URL-based read-before-publish updates, followed by surface-specific Artifact location and republish instructions.

Tool Description: WebFetch (concise) (249 tks) - Concise 工具 description for WebFetch co URL fetching, private URL limitations, redirects, and caching.

Tool Description: WebSearch (concise) (88 tks) - Describes the concise WebSearch 工具 with US-only results, current-month guidance, domain filters, and required sources.

工具 Description: Write (read existing file first) (121 tks) - 工具 description for Write in en where existing files must be read before o.

Tool Description: list_connected_browsers (browser picker guidance) (234 tks) - Appended to the list_connected_browsers 工具 description: when to ask the user to pick among several connected browsers, how to label the options, and to call select_browser or switch_browser rather than choosing one unilaterally.

工具 Parameter: Artifact URL guidance (app wording) (284 tks) - App-worded Artifact url parameter: an existing artifact's claude.ai link to update in place or act on, with the requirement to read an artifact this con has neither read nor published before publishing to it and to merge into a refusal's li.

Tool Parameter: Artifact URL guidance (333 tks) - Explains when the Artifact url parameter must name an existing owned artifact's claude.ai link, when to omit it for new publishes or same-con redeploys, and that publishing to an unread artifact requires reading it first and merging into any refused li.

工具 Parameter: Artifact auto-open timing (app wording) (163 tks) - Describes the auto_open parameter on a type_url Artifact create in third-person app wording: when Claude passes after_first_write so the person does not first see an empty Artifact, and when it omits the parameter so the Artifact opens on creation.

工具 Parameter: Artifact auto-open timing (179 tks) - Describes the auto_open parameter on a type_url Artifact create: when to pass after_first_write so the user does not first see an empty Artifact, and when to omit it so the Artifact opens on creation.

工具 Parameter: Artifact comment actions guidance (688 tks) - Describes Artifact action parameter semantics for reading, replying to, resolving, and paginating comment threads.

工具 Parameter: Artifact 数据库 access level (181 tks) - Describes the read_db and write_db as_le for checking artifact access rules as a view, interact, or admin user, which only narrows access, keeps the caller's identity, and makes refused writes read as not found and refused reads as empty.

工具 Parameter: Artifact 数据库 batch writes (347 tks) - Defines the Artifact writes parameter for batched 数据库 mutations, including per-document pins and atomic or sequential commit behavior.

工具 Parameter: Artifact 数据库 precondition (182 tks) - Defines the Artifact database if_ precondition, required on every write to an existing document and omitted only when creating one, with the write refused until the document has been read and batch entries pinned indi.

工具 Parameter: Artifact force o guidance (170 tks) - Restricts Artifact force o to explicit user-appro of a specific newer and requires merge-first conflict handling.

工具 Parameter: Artifact force page o warning (55 tks) - Warns that forcing an Artifact publish discards the newer published page created by another session or an in-page save.

工具 Parameter: Artifact force supporting files preser (58 tks) - Clarifies that a forced Artifact publish preser files unless the publish replaces or explicitly removes them.

工具 Parameter: Artifact preview action (211 tks) - Describes the Artifact check 工具's preview action value, which renders one local HTML page file at chosen widths and themes and returns height-capped screenshots plus a layout and load problem checklist without uploading anything.

工具 Parameter: Artifact supporting files with cross-artifact sources (app wording) (248 tks) - App-worded Artifact files parameter guidance for local sources, typed sources, ser cross-artifact copies, update remo, and 源码- limits.

工具 Parameter: Artifact supporting files with cross-artifact sources (233 tks) - Describes Artifact supporting-file maps and lists, ser copies from other Artifact , update preser and removal, and local 源码 restrictions.

Tool Parameter: Artifact title (267 tks) - Describes concise Artifact titles, HTML title-tag precedence, Markdown filename identity, and conditional naming behavior for creation from an Artifact type.

工具 Parameter: Artifact unread path o acknowledgement (149 tks) - Defines o paths for an existing multi-file Artifact while preserving change-detection refusal for every other path.

工具 Parameter: Artifact watch actions guidance (552 tks) - Describes Artifact watch, unwatch, status, li subscriptions, comment notification behavior, and explicit resume_replies behavior.

工具 Parameter: Bash command description (311 tks) - Instructions for generating clear, concise command descriptions in acti for bash commands.

Tool Parameter: Bash run_in_background guidance (127 tks) - Explains Bash run_in_background behavior and that commands do not need a trailing ampersand.

Tool Parameter: Bash run_in_background note (107 tks) - Notes that Bash commands can use run_in_background when the result is not needed immediately.

工具 Parameter: Claude in Chrome Ja 代码 (103 tks) - Describes the Ja 代码 parameter for the Claude in Chrome Ja execution 工具.

Tool Parameter: Computer action (251 tks) - Action parameter options for the Chrome browser computer 工具.

工具 Parameter: Sandboxed command allowed domains (228 tks) - Declares additional 网络 hosts for per-command auto-mode appro rejecting host suggestions from untrusted content.

工具 Parameter: SendUserMessage attachments (75 tks) - Describes optional SendUserMessage attachments as local file paths or pre-resolved file objects.

Tool Parameter: matched ask rule (198 tks) - Describes metadata identifying a user-configured permissions.ask rule that forced a 工具 appro while preserving the 工具-authored decision reason.

Tool Parameter: set_cwd needs_trust directory (197 tks) - Describes the canonical target directory returned by a set_cwd needs_trust response, which the SDK host must show in a trust dialog and echo back on accept.

Skills

Built-in skill prompts for specialized tasks.

Skill: /代码-review efficiency dimension (106 tks) - 代码-review pass that surfaces wasted effort the diff adds — duplicate computation or I/O, a serialization, large scopes held by closures — and points to the cheaper option.

Skill: /doctor prompt-audit configuration scope (532 tks) - Default arguments /doctor prompt-audit hands to the bundled claude-api skill, scoping the prompt audit to Claude 代码 configuration files that load in this project, excluding settings files and secrets, limiting edits outside the project, and treating audited files as data.

Skill: /doctor slash command description (372 tks) - Trigger description for the /doctor slash command co setup health, unused extensions, memory cleanup, hooks, updates, and permission prompts.

Skill: /doctor slash command (16135 tks) - Diagnostic workflow for auditing and fixing Claude 代码 installation health, unused context, local memory duplication, hooks, currency, and permission prompts.

Skill: /explain-usage measured usage slash command (986 tks) - Hands Claude the con's measured token usage as JSON with group definitions and caveats, asking for one simple chart of effecti by group and a plain-language explanation.

Skill: /explain-usage slash command (449 tks) - Analyzes the current session transcript into cost-weighted token usage groups, charts the results, and explains them in plain language.

Skill: /init CLAUDE.md and skill setup (new ) (7828 tks) - A comprehensi flow for setting up CLAUDE.md and related skills/hooks in the current repository, including codebase exploration, user interviews, and iterati refinement.

Skill: /insights report output (293 tks) - Formats and displays the insights usage report results after the user runs the /insights slash command.

Skill: /loop cloud-first scheduling offer (730 tks) - Decision tree for offering cloud-based scheduling before falling back to local session loops in the /loop command.

Skill: /loop local runtime note (96 tks) - Conditional /loop confirmation note explaining that local loops run only until the current session closes.

Skill: /loop self-pacing mode (re-arm decision) (412 tks) - Re-arm decision step for self-paced /loop iterations: call ScheduleWakeup with a fallback-heartbeat or cadence delay, a short reason, the original /loop input as the prompt, and a noop flag, or stop the loop instead.

Skill: /loop self-pacing mode (738 tks) - Instructs Claude how to self-pace a recurring loop by arming event monitors as primary wake signals and scheduling fallback heartbeat delays between iterations.

Skill: /loop slash command (dynamic mode) (977 tks) - Parses user input into an interval and prompt for scheduling recurring or dynamically self-paced loop executions.

Skill: /stuck (background-daemon diagnostics) (273 tks) - The background-daemon troubleshooting section of the /stuck skill.

Skill: /stuck slash command (964 tks) - Diagnoses frozen or slow Claude 代码 sessions.

Skill: Artifact PR review description (composed publish flow) (178 tks) - Trigger description for the composed PR review briefing Artifact flow with republish-only updates.

Skill: Artifact components (2455 tks) - Catalogs reusable Artifact components and supplies the exact scripts, styles, tokens, markup contract, and composition limits for embedding interacti blocks.

Skill: Artifact data table (2016 tks) - Skill instructions for creating interacti artifacts with sortable, filterable tabular data, safe JSON embedding, theming, and template-preser requirements.

Skill: Artifact diagramming (1352 tks) - Provides Artifact diagramming guidance, including when diagrams add value and how to build legible theme-aware inline SVG.

Skill: Artifact document (2055 tks) - Skill instructions for creating li document artifacts with structured prose, status metadata, commenting, and template-preser requirements.

Skill: Artifact explainer (2024 tks) - Instructions for creating explainer artifacts with numbered-step or section walkthrough structures from the built-in template.

Skill: Artifact report (1930 tks) - Instructions for creating long-form report artifacts with a masthead, table of contents, structured sections, and optional appendix from the built-in template.

Skill: Artifact whiteboard when-to-use guidance (182 tks) - When-to-use guidance for proactively offering a whiteboard during architecture, multi-component planning, or repeated clarification while limiting the offer to once per session.

Skill: Claude 代码 configuration guide (2462 tks) - Skill instructions for answering Claude 代码 configuration questions by checking the running build, bundled references, and current documentation.

Skill: Claude Test sign-in (371 tks) - Claude Test skill that signs the site's dedicated test member in before a run and saves the browser session, so specs needing an account start signed in.

Skill: Claude guide una reference fallback (344 tks) - Provides the Claude guide skill with an embedded li reference and fallback instructions when its reference files cannot be written to disk.

Skill: Code Review (Angle B — remo auditor) (94 tks) - 代码-review finder angle that, for each deleted or rewritten line, names the behavior it guaranteed and confirms the new 代码 still guarantees it.

Skill: Code Review (Angle C — cross-file tracer) (88 tks) - 代码-review finder angle that follows each changed function out to its callers, checking the diff hasn't broken a call-site contract.

Skill: Code Review (Angle D — language-pitfall specialist) (101 tks) - 代码-review finder angle that hunts for the well-known traps of the diff's language or 框架.

Skill: Code Review (Angle E — wrapper/proxy correctness) (126 tks) - 代码-review finder angle for wrapping types (caches, proxies, decorators), checking every method forwards faithfully to the wrapped object.

Skill: Code Review (Output — findings JSON array) (343 tks) - Defines the 代码-review skill's result shape: a JSON array of findings carrying file, line, summary, and failure_scenario.

Skill: Code Review (Phase 0 — gather the diff) (135 tks) - Opening step of the 代码-review skill: assemble the unified diff to review with git diff.

Skill: Code Review (Phase 2 — , 3-state) (125 tks) - Precision-tier step: run one per candidate finding, each CONFIRMED, PLAUSIBLE, or REFUTED.

Skill: Code Review (Phase 2 — , recall-biased) (137 tks) - Recall-tier step: one per candidate finding, biased toward keeping anything plausible.

Skill: Code Review (Phase 3 — sweep for gaps) (131 tks) - Final 代码-review sweep: a clean-slate re re-reads the diff to catch defects the earlier passes missed.

Skill: Code Review (con dimension) (0 tks) - 代码-review dimension: flag diff lines that break a rule stated in an applicable CLAUDE.md (user, repo-root, or ancestor-directory), quoting the exact rule and offending line, and emit nothing when no CLAUDE.md governs the change.

Skill: 代码 Review correctness finder angles (414 tks) - Shared /代码-review finder-angle block co line-by-line diff scanning, remo auditing, and cross-file tracing.

Skill: 代码 Review inline medium/high template (652 tks) - Template for medium and high inline 代码-review prompts that run eight finder angles, deduplicate without , and enforce a minimum findings target.

Skill: Code Review inline xhigh mode (1234 tks) - Extra-high inline /代码-review prompt that runs ten finder angles, deduplicates without , sweeps for gaps, and returns up to fifteen findings.

Skill: 代码 Review low effort expanded-findings mode (810 tks) - Low-effort /代码-review prompt that reads the diff once, returns up to eight hunk- findings, and targets at least min(files_changed, 4) genuine findings.

Skill: Computer Use MCP (1731 tks) - Instructions for using computer-use MCP tools including 工具 selection tiers, app access tiers, link safety, and financial action restrictions.

Skill: Data description (496 tks) - Trigger description for the Data skill co charts, graphs, dashboards, palettes, stat tiles, and related requests.

Skill: Data (2901 tks) - Skill instructions for producing accessible, brand-neutral charts, graphs, dashboards, and data using a method.

Skill: Debugging (653 tks) - Instructions for debugging an issue that the user is encountering in the Claude 代码 session.

Skill: Dynamic pacing loop execution (re-arm decision) (411 tks) - Re-arm decision step for dynamic-pacing /loop ticks: call ScheduleWakeup with a fallback-heartbeat or obser delay, a short reason, the dynamic-mode sentinel prompt, and a noop flag, or stop the loop instead.

Skill: Dynamic pacing loop execution (664 tks) - Step-by-step instructions for executing a dynamic pacing loop that runs tasks, arms and re-arms event monitors, schedules fallback heartbeat ticks, and handles task notifications.

Skill: Generate permission allowlist from transcripts (2408 tks) - Analyzes session transcripts to extract frequently used read-only 工具-call patterns and adds them to the project's .claude/settings.json permission allowlist to reduce permission prompts.

Skill: Import to Claude Code (418 tks) - Generated SKILL.md instructing Claude to finish importing lefto config that claude import could not map automatically.

Skill: Multiplayer whiteboard description (211 tks) - Trigger description for the multiplayer Whiteboard skill, co creation of a li canvas with synchronized strokes, cursors, session presence, pasted images, and Claude drawing.

Skill: Plugin e interview (5128 tks) - Guided interview for creating Claude plugin e under evals/ with gated inputs, graders, calibration, and cost checks.

Skill: Prototype description (151 tks) - Trigger description for the Prototype skill, co working proof-of-concept Artifacts and optional in-place feature demonstrations.

Skill: Prototype runtime capabilities guidance (242 tks) - Explains when prototype Artifacts should use user-granted runtime capabilities and requires loading the artifact-capabilities skill before relying on live data or actions.

Skill: Run CLI tool example (648 tks) - Example file for the Run app skill showing how to document building, in, and testing a CLI 工具.

Skill: Run TUI interacti app example (1292 tks) - Example file for the Run app skill showing how to drive an interacti app with tmux, readiness polling, pane capture, key references, and cleanup.

Skill: Run app (1315 tks) - Skill for launching and driving the current project's app through its real runtime surface using project-specific run skills or fallback patterns.

Skill: Run browser-driven web app example (1413 tks) - Example file for the Run app skill showing how to start a web de, drive it with chromium-cli, capture screenshots, and document app-specific gotchas.

Skill: Run library SDK example (835 tks) - Example file for the Run app skill showing how to document building, testing, and smoke-checking a library or SDK at its public package boundary.

Skill: Run skill template (1596 tks) - Template file for the Run skill generator showing the frontmatter and section structure for a project-specific run skill.

Skill: Run web server API example (1364 tks) - Example file for the Run app skill showing how to document a 服务端 or API lifecycle with background launch, readiness checks, curl , and shutdown.

Skill: Schedule recurring cron and run immediately (271 tks) - Converts an interval to a cron expression, schedules a recurring task via the cron creation 工具, confirms to the user, and immediately executes the task without waiting for the first cron fire.

Skill: Setup Cowork role selection (475 tks) - First step of the setup-cowork skill that explains Cowork, asks for the user's role, calls ShowOnboardingRolePicker, or falls back to a plain-text role list.

Skill: Setup Cowork (3464 tks) - Guided Cowork setup flow that helps the user pick a role, install matching plugins, connect tools, try a skill, configure writing voice, and wrap up.

Skill: Team onboarding guide (521 tks) - Template for onboarding a new teammate to a team's Claude 代码 setup, walking them through usage stats, setup checklists, MCP servers, skills, and team tips in a warm con style.

Skill: Update Claude Code Config (1726 tks) - Skill for modifying Claude 代码 configuration file (settings.json).

Skill: Update config description (170 tks) - Update-config skill description (settings.json hooks, perms, env).

Skill: Update config settings file locations (1445 tks) - Where Claude 代码 stores settings.json across scopes.

Skill: CLI changes (example for skill) (772 tks) - Example workflow for a CLI change, as part of the skill.

Skill: 服务端/API changes (example for skill) (793 tks) - Example workflow for a 服务端/API change, as part of the skill.

Skill: Whiteboard description (228 tks) - Trigger description for the Whiteboard skill, co creation of a shared sketch canvas for collaborati, design discussion, and planning.

Skill: Workflow authoring reference (6322 tks) - Skill reference for authoring Workflow scripts, co script persistence, metadata, agent hooks, isolation, concurrency, budgeting, quality patterns, and resume behavior.

Skill: update-config (7-step flow) (1755 tks) - A skill that guides Claude through a 7-step process to construct and hooks for Claude 代码, ensuring they work correctly in the user's specific project en.

项目信息

开发语言：JavaScript

Star 数：12832

技术要点

该项目属于安全软件方向的开源资源，代码结构清晰，适合安全研究人员学习参考。项目围绕All parts of Claude Code's system prompt, 27 builtin 工具 descriptions, sub agent prompts (Plan/Explore/Task), utility prompts (CLAUDE.md, compact, statusline, magic docs, WebFetch, Bash cmd, security review, agent creation). Updated for each Claude 代码 version.展开，提供了可运行的实现，对理解相关安全机制、工具原理与攻防技术具有较高的研究价值。

从技术实现角度看，此类项目的核心在于对底层原理的深入理解与工程化落地。开发者在研究过程中可以结合官方文档与社区讨论，梳理项目的初始化流程、核心模块职责与数据流转，逐步掌握其中的技术细节与设计思路。

使用与学习建议

1. 下载源码压缩包，解压到本地环境。

2. 根据项目类型准备对应的运行环境与依赖。

3. 按照 README 中的步骤完成配置与部署。

4. 在隔离的测试环境中运行验证，避免影响生产系统。

免责声明

本资源来源于开源社区，仅供安全研究、技术学习与合法授权范围内的攻防演练使用，请勿用于任何非法或侵害他人权益的用途。使用本项目产生的一切后果由使用者自行承担。

登录后才能参与讨论

下载链接

会员才能下载

登录后开通会员

同分类推荐

暂无同类资源

资源标题	claude-code-system-prompts - All parts of Claude Code's system prompt, 27 built
所属分类	安全软件
发布作者	admin
发布时间	2026-10-05 22:28:05
下载权限	免费下载
附件数量	2 个下载链接
浏览次数	537
讨论数量	0
主题状态	正常讨论中