Skip to content

Latest commit

 

History

History
40 lines (30 loc) · 2.41 KB

File metadata and controls

40 lines (30 loc) · 2.41 KB

学习资源清单(2026-07 调研)

必读教程(按顺序)

  1. How to Build an Agent — Thorsten Ball(Amp) https://ampcode.com/notes/how-to-build-an-agent M1 的直接蓝本:Go 写的 ~400 行 code-editing agent(read_file / list_files / edit_file 三个工具)。名言:"It's an LLM, a loop, and enough tokens." Python 改写版:https://medium.com/@jbrathnayake98/how-to-build-an-agent-by-thorsten-ball-python-version-ebbabb8665f6

  2. how to build a coding agent(免费 workshop) — Geoffrey Huntley https://ghuntley.com/agent/ | 代码:https://github.com/ghuntley/how-to-build-a-coding-agent 从 chatbot 逐步加文件读写、bash、搜索工具的完整 workshop,和我们的里程碑划分思路一致。

  3. Build a Coding Agent from Scratch: The Complete Python Tutorial — Sid Bharath https://sidbharath.com/blog/build-a-coding-agent-python-tutorial/

Claude Code 架构分析(做完 M1 再读)

  1. 官方文档:How the agent loop works https://code.claude.com/docs/en/agent-sdk/agent-loop
  2. Dive into Claude Code(VILA-Lab,系统性学术分析,含 arXiv 论文) https://github.com/VILA-Lab/Dive-into-Claude-Codehttps://arxiv.org/html/2604.14228v1
  3. Tracing Claude Code's LLM Traffic — George Sung https://medium.com/@georgesung/tracing-claude-codes-llm-traffic-agentic-loop-sub-agents-tool-use-prompts-7796941806f5 抓包分析真实请求:agent 循环、子 agent、system prompt 长什么样。
  4. Claude Code: Behind-the-scenes of the master agent loop — PromptLayer https://blog.promptlayer.com/claude-code-behind-the-scenes-of-the-master-agent-loop/
  5. Claude Code Architecture Explained(DEV Community) https://dev.to/brooks_wilson_36fbefbbae4/claude-code-architecture-explained-agent-loop-tool-system-and-permission-model-rust-rewrite-41b2

参考实现(对照我们的代码看)

  1. claw-code — ~5000 行 Python 的 Claude Code 干净重写,架构映射清晰(GitHub 搜 claw-code)
  2. nano-claude-codehttps://github.com/SafeRL-Lab/nano-claude-code 极简可用版
  3. 开源复现合集https://github.com/chauncygu/collection-claude-code-source-code

API 文档

  1. Anthropic Tool Use 官方文档:https://docs.anthropic.com/en/docs/build-with-claude/tool-use
  2. Anthropic《Building Effective Agents》:https://www.anthropic.com/research/building-effective-agents