Minbook
KO
Claude Code Anatomy
Agents & Architecture Active 5 posts

Claude Code Anatomy

Anatomy of Claude Code's production agent system, exposed by the 2026-03 npm sourcemap leak — execution loop, tool system, memory, and unreleased features.

About this series

A March 2026 npm sourcemap leak exposed the internals of Claude Code in full. Four-stage execution loop, seven modes, tool system, memory layers, unreleased features — production agent design has rarely been this visible.

Useful for builders creating their own agents, Claude Code power users wanting to know how it works, and analysts treating AI tool architecture as study material.

Read in order: architecture → tools → memory → hidden features. As you go, separate "what we can borrow" from "what only makes sense inside Claude" — that's how you end up with patterns transferable to your own system.

5 episodes

  1. 01
    Anchor post

    What the Claude Code Leak Revealed: Anatomy of an AI Agent

    The March 31, 2026 npm sourcemap incident revealed Claude Code internals. 4-phase execution, 7 modes, and the 11-step Agent Loop analyzed.

  2. 02

    Agent System Design Canvas — 12 Production Patterns Proven by the Claude Code Leak

    6-Layer agent system design canvas and 8+4 core patterns from the Claude Code source leak (512K lines TS). From the Circuit Breaker that stopped 250K/day wasted API calls with 3 lines, to 23-step bash AST security.

  3. 03

    52 Tools, 23-Step Security: Inside an Agent's Tool System

    52 built-in tools' common interface, 10-step execution pipeline, safe=parallel/unsafe=sequential concurrency model, 5-stage permission pipeline, and 888KB Tree-sitter 23-step bash security.

  4. 04

    The Memory System I Built Looked Like Claude Code's Internal Design

    CC internal Memory System (4-type persistent memory, Auto-Dream, Auto-Compact) vs independently built 3-layer system (documents→index→semantic search). validate_placement() is the differentiator CC doesn't have.

  5. 05

    KAIROS, Auto-Dream, Coordinator: What Unreleased Features Reveal About AI's Future

    44 feature flags, 20 externally inactive. KAIROS, Auto-Dream, UltraPlan, Coordinator, Bridge, Daemon, UDS Inbox, Buddy, plus anti-distillation and undercover mode.