Learning Lab

This repository is a personal learning lab for agent engineering, LLM application frameworks, and market research tooling.

Areas

• claude-code/ - Claude Code style harness engineering course material, runnable Python agents, skills, tests, and the learning website.
• langchain/ - reserved workspace for LangChain and LangGraph experiments.
• github-ai-learning/ - Obsidian-friendly study notes for GitHub AI projects, including source reading, reproduction notes, and reusable takeaways.
• stock-research/ - A-share research tooling with akshare-one-mcp, repeatable Python reports, and risk checks.
• shared/ - cross-topic utilities, templates, and notes that do not belong to a single learning area.
• docs/ - repository-level learning roadmap and index.

Quick Start

Claude Code course:

python claude-code/s01_agent_loop/code.py

A-share research script:

uv run --with akshare --with pandas --with numpy python stock-research/tools/a_share_researcher.py 600519 --start 20240101 --end 20260529

Web course site:

cd claude-code/web
npm run dev

Repository Policy

Track source code, prompts, templates, and learning notes. Ignore generated reports, market data exports, caches, IDE state, and runtime mailboxes.