daiuuuuuuuuuuuuu/visual-proxy
Claude Code skill — 给无视觉能力的 LLM（DeepSeek/o1/o3）外挂看图能力。Vision proxy for text-only LLMs, supports OpenAI/Anthropic/DashScope/Qwen3-VL.
Claude Code skill — 给无视觉能力的 LLM（DeepSeek/o1/o3）外挂看图能力。Vision proxy for text-only LLMs, supports OpenAI/Anthropic/DashScope/Qwen3-VL.
npx add-skill daiuuuuuuuuuuuuu/visual-proxyClaude Code skill — 给无视觉能力的 LLM（DeepSeek/o1/o3）外挂看图能力。Vision proxy for text-only LLMs, supports OpenAI/Anthropic/DashScope/Qwen3-VL.
Exercise: Get started using GitHub Copilot
Turn vibe coding into production-ready software. 20 years of engineering experience distilled into AI agent skills.
Turn any concept into a self-contained interactive HTML study page: a parable teaches it indirectly, then the Feynman technique makes you explain it back — with answers locked until you try. A Claude / openclawmp skill (ships Chinese-first).
Codex skill for scoped product UI work with preserved behavior and evidence-checked handoff.
Explore Claude Code source to study AI coding tool design, architecture, prompt engineering, and engineering practices
Build a pure-Python CLI agent framework for LLM workflows, with provider-agnostic design, clear code, and async multi-agent support