AI

Claude code

An index for the Claude Code source analysis series, organized around architecture, core mechanisms, tools, extensibility, and collaboration.

中文 English 日本語
返回 AI / Claude code

阅读路径

16 篇文章

Claude code

An index for the Claude Code source analysis series, organized around architecture, core mechanisms, tools, extensibility, and collaboration.

开始阅读

目录

按顺序读下去，会从模型原理一路走到 Agent 与 Harness 的工程视角。

  1. 00 Guide to the Claude Code Source Analysis Series 导读 An index for the Claude Code source analysis series, organized around architecture, core mechanisms, tools, extensibility, and collaboration.
  2. 01 Claude Code Source Analysis Series, Chapter 1: Architecture A high-level tour of Claude Code's runtime architecture, capability layers, and core modules.
  3. 02 Claude Code Source Analysis Series, Chapter 2: The ReAct Main Loop How Claude Code implements its ReAct loop and keeps the model moving through multi-step work.
  4. 03 Claude Code Source Analysis Series, Chapter 3: Prompt Construction How Claude Code assembles its system prompt, memory, and runtime context on each turn.
  5. 04 Claude Code Source Analysis Series, Chapter 4: Context Management How Claude Code manages context, compresses history, and stays coherent during long-running tasks.
  6. 05 Context Governance for Coding Agents A broader industry view of context management and how Claude Code compares with other agent systems.
  7. 06 Claude Code Source Analysis Series, Chapter 6: Tools Overview A breakdown of Claude Code's tool system, including tool protocols, permissions, and execution flow.
  8. 07 Claude Code Source Analysis Series, Chapter 7: File Tools How Claude Code reads, edits, and writes files safely inside a real codebase.
  9. 08 Claude Code Source Analysis Series, Chapter 8: Terminal Tools How Claude Code runs terminal commands, enforces permissions, and feeds results back into the loop.
  10. 09 Claude Code Source Analysis Series, Chapter 9: Task Management How Claude Code tracks todos, plans, and long-running task state through workflow tools.
  11. 10 Claude Code Source Analysis Series, Chapter 10: MCP How Claude Code integrates MCP servers and turns external tools and resources into first-class runtime capabilities.
  12. 11 Claude Code Source Analysis Series, Chapter 11: Skills How Claude Code packages task experience into reusable skills and loads them at runtime.
  13. 12 Claude Code Source Analysis Series, Chapter 12: Agent Collaboration How Claude Code coordinates multiple agents, including subagents, tasks, messaging, and permission boundaries.
  14. 13 Claude Code Source Analysis Series, Chapter 13: Industry Patterns for Agent Collaboration (Optional) A cross-industry comparison of multi-agent collaboration patterns, and how different systems split work and converge on results.
  15. 14 Claude Code Source Analysis Series, Chapter 14: Plan How Claude Code turns Plan Mode into a runtime mechanism with permission boundaries, approvals, persistence, and verification.
  16. 15 Claude Code Source Analysis Series, Chapter 15: Planning Patterns Across Agent Systems (Optional) How planning in modern agent systems evolves from prompt advice into an execution control plane.