Claude Code System Prompts

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

课时

• Prompts
  • Agent Prompts
    • Sub-agents
      • Agent Prompt: Explore
      • Agent Prompt: Plan mode (enhanced)
    • Creation Assistants
      • Agent Prompt: CLAUDE.md creation
      • Agent Prompt: Status line setup
    • Slash Commands
      • Agent Prompt: /batch slash command
      • Agent Prompt: /code-review GitLab comment posting
      • Agent Prompt: /code-review inline gap sweep phase
      • Agent Prompt: /code-review minimal mode
      • Agent Prompt: /code-review part 1 base finder angles
      • Agent Prompt: /code-review part 10 ReportFindings output format
      • Agent Prompt: /code-review part 2 low effort minimum findings mode
      • Agent Prompt: /code-review part 2 low effort mode
      • Agent Prompt: /code-review part 3 extra-high and maximum effort modes
      • Agent Prompt: /code-review part 4 three-state verification phase
      • Agent Prompt: /code-review part 5 recall-biased verification phase
      • Agent Prompt: /code-review part 6 medium effort mode
      • Agent Prompt: /code-review part 7 high effort mode
      • Agent Prompt: /code-review part 8 GitHub comment posting
      • Agent Prompt: /code-review part 9 fix application
      • Agent Prompt: /code-review unavailable-agent inline mode
      • Agent Prompt: /rename auto-generate session name
      • Agent Prompt: /schedule slash command
      • Agent Prompt: /security-review slash command
      • Agent Prompt: /simplify slash command
      • Agent Prompt: /simplify unavailable-agent inline mode
      • Agent Prompt: /ultrareview GitHub comment poster
      • Agent Prompt: Artifact type creation slash command
    • Utilities
      • Agent Prompt: Agent Hook
      • Agent Prompt: Artifact comment thread analyst
      • Agent Prompt: Artifact editor thread follow-up
      • Agent Prompt: Auto mode rule reviewer
      • Agent Prompt: Away summary generation
      • Agent Prompt: Background agent state classifier
      • Agent Prompt: Background job agent instructions
      • Agent Prompt: Claude Code guide
      • Agent Prompt: Claude guide agent
      • Agent Prompt: Coding session title generator
      • Agent Prompt: Conversation summarization
      • Agent Prompt: Coordinator worker instructions
      • Agent Prompt: Determine which memory files to attach
      • Agent Prompt: Dream memory consolidation
      • Agent Prompt: General purpose agent
      • Agent Prompt: General purpose
      • Agent Prompt: General task agent
      • Agent Prompt: Hook condition evaluator (stop)
      • Agent Prompt: Hook condition evaluator
      • Agent Prompt: Inherited context for worktree sub-agent
      • Agent Prompt: Onboarding guide draft share link workflow
      • Agent Prompt: Onboarding guide generator
      • Agent Prompt: PR follow-up cron
      • Agent Prompt: Project thread status card classifier
      • Agent Prompt: Prompt Suggestion Generator v2
      • Agent Prompt: Pull request creation
      • Agent Prompt: Quick PR creation
      • Agent Prompt: Quick git commit
      • Agent Prompt: Read-only search agent
      • Agent Prompt: Remote machine auto mode rules
      • Agent Prompt: Schedule action selection
      • Agent Prompt: Schedule local MCP server limitation
      • Agent Prompt: Security monitor Claude Tag connector writes
      • Agent Prompt: Security monitor for autonomous agent actions (first part)
      • Agent Prompt: Security monitor forwarded user turns
      • Agent Prompt: Security monitor host-context line guidance
      • Agent Prompt: Session title and branch generation
      • Agent Prompt: Session transcript chunk summary
      • Agent Prompt: Summarization no-tools guard
      • Agent Prompt: Web fetch agent usage guidance
      • Agent Prompt: Web reading specialist
      • Agent Prompt: Worker fork
      • Agent Prompt: Workflow script plain text return note
      • Agent Prompt: Workflow script structured return note
      • Agent Prompt: Workflow subagent plain text output
      • Agent Prompt: Workflow subagent structured output
  • Data
    • Data: /auto-mode-setup usage
    • Data: Artifact MCP connector guidance
    • Data: Artifact connector call observation requirement
    • Data: Artifact host MCP server guidance
    • Data: Artifact runtime capability declarations
    • Data: Auto-compact inputs changed event schema
    • Data: Background tasks changed event schema
    • Data: Claude API reference — Go
    • Data: Claude API reference — PHP
    • Data: Claude API reference — Ruby
    • Data: Claude Code agent proxy troubleshooting guide
    • Data: Claude Code gateway customer-routed inference protocol
    • Data: Claude Code gateway protocol
    • Data: Claude Code live documentation sources
    • Data: Claude Code recent changes reference
    • Data: Claude Platform on AWS reference
    • Data: Claude Tag (Claude in Slack) reference
    • Data: Claude gateway landing page
    • Data: Code change published event schema
    • Data: Command plugin source command field
    • Data: Cowork plugin MCP discovery and connection
    • Data: Cowork plugin examples
    • Data: Cross-session inbound setting
    • Data: Data visualization anti-patterns
    • Data: Data visualization choosing a form
    • Data: Data visualization components
    • Data: Data visualization interaction
    • Data: Data visualization marks and anatomy
    • Data: Dialog expiry setting
    • Data: DirectoryAdded hook description
    • Data: Files API reference — C#
    • Data: Files API reference — Go
    • Data: Files API reference — Java
    • Data: Files API reference — PHP
    • Data: Files API reference — Python
    • Data: Files API reference — TypeScript
    • Data: GitHub Actions workflow for @claude mentions
    • Data: GitHub App installation PR description
    • Data: Hook classifier context field
    • Data: Interrupt cancel queued parameter
    • Data: Interrupt receipt cancelled field
    • Data: Interrupt receipt still queued field
    • Data: Knowledge MCP search strategies
    • Data: MCP HTTP client capability validation rationale
    • Data: MCP HTTP parameter header validation rationale
    • Data: MCP HTTP standard header validation rationale
    • Data: Managed Agents outcomes
    • Data: Managed settings helper onFailure field
    • Data: Managed settings helper retries field
    • Data: Message Batches API reference — C#
    • Data: Message Batches API reference — PHP
    • Data: Message Batches API reference — Python
    • Data: Message Batches API — TypeScript
    • Data: Peer sender display name field
    • Data: Platform availability
    • Data: Published model catalog seed guidance
    • Data: Query result pending command count
    • Data: Rate limit unified windows
    • Data: Review upload excluded changes error
    • Data: Rewind files skippedLinks field
    • Data: SDK MCP server errors field
    • Data: SDK MCP server manifests field
    • Data: SDK Remote Control availability field
    • Data: SDK assistant user message UUID field
    • Data: SDK assistant user message UUIDs field
    • Data: SDK cloud session init snapshot field
    • Data: SDK control cancel request schema
    • Data: SDK error result user message UUID field
    • Data: SDK error result user message UUIDs field
    • Data: SDK footer indicator schema
    • Data: SDK initialize plugins parameter
    • Data: SDK partial assistant user message UUID field
    • Data: SDK per-task stop affordance field
    • Data: SDK plugin warnings field
    • Data: SDK protocol capabilities field
    • Data: SDK query result modelUsage field
    • Data: SDK register device hooks request schema
    • Data: SDK remote tool call request schema
    • Data: SDK request user dialog kind field
    • Data: SDK result message schema
    • Data: SDK set max thinking tokens request schema
    • Data: SDK set model system prompt field
    • Data: SDK subagent stats schema
    • Data: SDK workspace trust directory field
    • Data: Sandbox TLS termination setting
    • Data: Sandbox credential environment no-match setting
    • Data: Sandbox credential file mask claims setting
    • Data: Sandbox credential mask empty injectHosts warning
    • Data: Sandbox filesystem disabled setting
    • Data: Sandbox network domain spelling warning
    • Data: Self-hosted runner command help
    • Data: Self-hosted runner decode-token command help
    • Data: Self-hosted runner deferred shutdown timing notice
    • Data: Self-hosted runner orchestrator command help
    • Data: Streaming reference — C#
    • Data: Streaming reference — Go
    • Data: Streaming reference — Java
    • Data: Streaming reference — PHP
    • Data: Streaming reference — Python
    • Data: Streaming reference — Ruby
    • Data: Streaming reference — TypeScript
    • Data: Thin-client diff dialog schema
    • Data: Token counting reference
    • Data: Tool use display metadata field
    • Data: Tool use reference — C#
    • Data: Tool use reference — PHP
    • Data: Tool use reference — Ruby
    • Data: Upload device hook template request
    • Data: VCS state changed branch field
    • Data: VCS state changed event schema
    • Data: allowedMcpServers setting
    • Data: syncClaudeAiPlugins setting
    • Data: syncClaudeAiSkills setting
  • System Prompt
    • System Prompt: Act when ready
    • System Prompt: Action safety and truthful reporting
    • System Prompt: Advisor tool instructions
    • System Prompt: Agent Summary Generation
    • System Prompt: Agent thread notes
    • System Prompt: Artifact comment decision reformat retry
    • System Prompt: Artifact comment edit composer
    • System Prompt: Artifact comment fast acknowledgement selection
    • System Prompt: Artifact comment fast acknowledgement
    • System Prompt: Artifact comment list framing
    • System Prompt: Artifact comment presence state guidance
    • System Prompt: Artifact comment reply composer
    • System Prompt: Artifact comment result guidance
    • System Prompt: Artifact comment thread framing
    • System Prompt: Artifact comment thread triage
    • System Prompt: Auto memory durable lesson instructions
    • System Prompt: Auto mode setup proposal generator
    • System Prompt: Auto mode
    • System Prompt: Autonomous loop check
    • System Prompt: Autonomous loop notification guidance
    • System Prompt: Autonomous loop tick (dynamic pacing)
    • System Prompt: Autonomous loop tick
    • System Prompt: Autonomous operation guidelines
    • System Prompt: Background session instructions
    • System Prompt: Background session worktree persistence guidance
    • System Prompt: Background worktree isolation guidance
    • System Prompt: Censoring assistance with malicious activities
    • System Prompt: Chrome browser MCP tools
    • System Prompt: Claude Fable 5 model identity
    • System Prompt: Claude Fable 5.1 model identity
    • System Prompt: Claude in Chrome browser automation
    • System Prompt: Claude in Chrome browser selection instructions
    • System Prompt: Combined memory index pointer instructions
    • System Prompt: Comment what and task context avoidance
    • System Prompt: Comment why-only guidance
  • Skills
    • Skill: /code-review efficiency dimension
    • Skill: /doctor slash command description
    • Skill: /doctor slash command
    • Skill: /explain-usage slash command
    • Skill: /init CLAUDE.md and skill setup (new version)
    • Skill: /insights report output
    • Skill: /loop cloud-first scheduling offer
    • Skill: /loop local runtime note
    • Skill: /loop self-pacing mode
    • Skill: /loop slash command (dynamic mode)
    • Skill: /stuck (background-daemon diagnostics)
    • Skill: /stuck slash command
    • Skill: Artifact PR review description (composed publish flow)
    • Skill: Artifact components
    • Skill: Artifact dashboard
    • Skill: Artifact data table
    • Skill: Artifact diagramming
    • Skill: Artifact document
    • Skill: Artifact explainer
    • Skill: Artifact report
    • Skill: Artifact whiteboard when-to-use guidance
    • Skill: Claude Code configuration guide
    • Skill: Claude guide unavailable reference fallback
    • Skill: Code Review (Angle B — removed-behavior auditor)
    • Skill: Code Review (Angle C — cross-file tracer)
    • Skill: Code Review (Angle D — language-pitfall specialist)
    • Skill: Code Review (Angle E — wrapper/proxy correctness)
    • Skill: Code Review (Output — findings JSON array)
    • Skill: Code Review (Phase 0 — gather the diff)
    • Skill: Code Review (Phase 2 — verify, 3-state)
    • Skill: Code Review (Phase 2 — verify, recall-biased)
    • Skill: Code Review (Phase 3 — sweep for gaps)
    • Skill: Code Review (conventions dimension)
    • Skill: Code Review correctness finder angles
    • Skill: Code Review inline medium/high template
    • Skill: Code Review inline xhigh mode
    • Skill: Code Review low effort expanded-findings mode
    • Skill: Computer Use MCP
    • Skill: Data Visualization description
    • Skill: Data Visualization
    • Skill: Debugging
    • Skill: Design description
    • Skill: Dynamic pacing loop execution
    • Skill: Generate permission allowlist from transcripts
    • Skill: Import to Claude Code
    • Skill: Multiplayer whiteboard description
    • Skill: Plugin authoring
    • Skill: Plugin eval authoring interview
    • Skill: Prototype description
    • Skill: Prototype runtime capabilities guidance
    • Skill: Run CLI tool example
    • Skill: Run TUI interactive terminal app example
    • Skill: Run app
    • Skill: Run browser-driven web app example
    • Skill: Run library SDK example
    • Skill: Run web server API example
    • Skill: Schedule recurring cron and run immediately
    • Skill: Setup Cowork role selection
    • Skill: Setup Cowork
    • Skill: Team onboarding guide
    • Skill: Update Claude Code Config
    • Skill: Update config description
    • Skill: Update config settings file locations
    • Skill: Verify CLI changes (example for Verify skill)
    • Skill: Verify server/API changes (example for Verify skill)
    • Skill: Whiteboard description
    • Skill: Workflow authoring reference
    • Skill: update-config (7-step verification flow)
• Autonomous loop check
• Claude Code System Prompts

开始学习 → Claude Code System Prompts