Claude Code System Prompts — Full Extraction

What's Included

Agent Prompts

Sub-agent system prompts for:

• Explore — Fast codebase exploration agent
• Plan — Software architecture planning agent
• Slash commands: /batch, /pr-comments, /review-pr, /schedule, /security-review
• Utility agents: memory consolidation, security monitoring, verification

System Prompts

Core behavioral instructions covering:

• Auto/learning/minimal/plan modes
• Tool usage policies and priorities
• Task execution principles
• Code safety and security guidelines
• Delegation and agent spawning patterns

Builtin Tool Descriptions

Complete documentation for 18+ tools:

• Read, Edit, Write — File operations
• Bash — Shell execution with sandbox policies
• Grep, Glob — Code search
• Agent — Sub-agent spawning
• WebFetch, WebSearch — Web access
• TodoWrite, TaskCreate — Task management
• Computer — Browser automation

System Reminders

~40 runtime context injections:

• File modification notifications
• IDE interaction context
• Hook feedback handling
• Plan mode activation
• Token/budget tracking
• Memory contents injection

Data Templates

Reference materials embedded in Claude Code:

• Agent SDK patterns (Python/TypeScript)
• Claude API references (C#, Go, Java, PHP, Python, Ruby, TypeScript)
• GitHub Actions workflow templates
• Prompt caching design patterns

Version Tracking

Every Claude Code release is tracked with a detailed changelog showing exactly what system prompt strings were added, modified, or removed. Star the repo to get notified of new versions.