WenyuChiou/awesome-agentic-ai-zh

git clone https://github.com/WenyuChiou/awesome-agentic-ai-zh.git

Usage

• •Start from stages/00-foundations.md (or resources/setup-guide.md if completely new to AI agents)
• •After shared Stage 0-2, pick Track A (CLI Power User, ~8-10 weeks) or Track B (Agent Builder, 16-22 weeks / 5-7 months)
• •Read docs/HOW_TO_USE.md first — exercise starter.py files are full solutions, not TODO skeletons (rename to starter_reference.py and rewrite yourself)
• •Branch extension routes by role: researcher / developer / teacher / knowledge worker / everyday user

Key features

• ✓8-stage learning roadmap with 2 tracks and 5 role-based extension branches, including two shared hubs (Stage 5 Claude Code ecosystem, Stage 8 Agent Interfaces)
• ✓240+ curated projects with star ratings, who it suits, what it teaches, and how to run (incl. local LLMs: Ollama, llama.cpp, LocalAI, MLX)
• ✓Catalog of 77+ MCP/Skill entries across 16 categories (resources/mcp-skills-catalog.md)
• ✓23 illustrative exercise folders with 70-150 line starters, dual-path Ollama/Anthropic SDK comparisons, and mock-based tests
• ✓Fully maintained in three languages (zh-TW / zh-CN / English), English is not a thin translation
• ✓Glossary of 30+ terms linking concepts to stages; automated quality check on newly added GitHub links (stars, license, archived status)

Best for

Chinese-speaking developers or self-learners (basic Python expected) who want a complete, staged path from LLM basics to building multi-agent systems and writing their own MCP servers.

Caveats

• ⚠Not a runnable MCP server — pure documentation / learning-roadmap / awesome-list repository (MIT)
• ⚠Time commitment is significant: Track A ~8-10 weeks, Track B realistically 5-7 months at 5-8 hr/week