kjfx/AI

This README shows how to install and configure Claude Code for users in mainland China to run Claude Code against cloud or local large models (Qwen, Deepseek, or Ollama-hosted models). Intended audience: developers or advanced users who need to bypass regional restrictions and route Claude Code to alternative model providers. Key points: - Prepare a working scientific/proxy network (examples: Clash Verge, v2rayN) or use proxy providers. - Install Node.js and Git. Add C:\Users\<user>\.local\bin to PATH. - Bypass initial onboarding by setting "hasCompletedOnboarding": true in .claude.json. - Configure environment variables (Windows setx) to point Claude Code to other providers: - For Qwen: set ANTHROPIC_API_KEY, ANTHROPIC_BASE_URL (Aliyun dashscope), ANTHROPIC_MODEL (qwen3.6-flash…). - For Deepseek: set corresponding ANTHROPIC_BASE_URL and model name. - Verify with echo %ANTHROPIC_*% (restart cmd after setx). - To use a local model via Ollama: run the model with ollama, then set ANTHROPIC_API_KEY="ollama", ANTHROPIC_BASE_URL="http://127.0.0.1:11434", and ANTHROPIC_MODEL to the running model name; restart cmd and run claude.

AI-related keywords found in this repo's description, topics, or README summary — grouped by category. Each badge links to the corresponding ranking detail page.