v0.6.3
Documentation: a complete setup and usage reference in the README covering every mode - CLI install, the full analyze + fix flag tables, provider configuration (anthropic/openai/gemini/openai-compatible with base-url + env keys, incl. reasoning-model endpoints like Nvidia NIM, Fireworks, Groq), per-host enable steps (Claude Code, Codex, Gemini, Cursor, plain LLM/API), and the J1-J6 output schema. No detector change.