Back to Library
Tool
3.1k

Claude Code System Prompts (Piebald)

Every Claude Code system prompt — main agent, subagent, skill prompts, memory, session — updated within minutes of each release.

#Claude Code#System Prompts#AI Agents#Prompts#Reverse Engineering
Repository

Overview

The most up-to-date repository of reconstructed Claude Code system prompts. Covers the main orchestrator agent, subagent worker forks, memory selection, skill invocation, security classifier, and session management prompts — all updated within minutes of new Claude Code releases.

Features

  • Main Agent Prompt: Full orchestrator prompt showing how Claude Code structures reasoning and tool calls.
  • Worker Fork Prompt: How subagent delegation is framed to prevent further spawning.
  • Memory Prompts: Context selection and memory consolidation prompt patterns.
  • Skill Prompts: How skill MD files are injected and scoped during execution.

Use Cases

  • Understanding exactly what Claude Code sees when it reads your CLAUDE.md.
  • Learning how to write CLAUDE.md and skills that align with the actual prompt structure.
  • Reverse-engineering agent behavior to design better multi-agent pipelines.

Technical Advantages

  • Real extracted prompts, not guesses — verified against observable Claude Code behavior.
  • Updated same-day as Claude Code releases, often within the hour.
  • Annotated with structural analysis and behavioral notes.