- Multi-provider LLM support: OpenAI, Mistral, Groq, OpenRouter, Google Gemini, DeepSeek, Together, Fireworks, Cerebras
[llm.providers.*]config blocks for multiple API keys in one file- Environment variable fallback per provider (
MISTRAL_API_KEY, etc.) organize-stars providerscommandorganize-stars plan --provider mistralCLI override[llm] preferences— natural-language hints for AI categorization- Fetches existing GitHub Star Lists before AI planning (reuses list names)
- Clear error when
categorization-plan.jsonis missing (instead of traceback) - Release workflow:
skip-existing: trueso re-running a tag does not fail
- Windows PATH note for
pip install(python -m stars_organizerfallback)
organize-stars init— interactive config setuporganize-stars status— config and resume state checkorganize-stars lists— show current Star Listsorganize-stars plan --categories— customcategories.tomlrulesorganize-stars plan --llm— optional AI categorization (uv sync --extra llm)organize-stars apply --resume— checkpoint resume via.organize-stars-state.json- Rich progress bar during apply
- PyPI publish workflow on version tags
- Removed Python → AI & LLM language bias
- Language hints used as tie-breakers only
- Pre-apply guard when plan exceeds GitHub's 32-list limit
organize-stars plan— fetch stars and build categorization planorganize-stars apply— create Star Lists and assign repos- Heuristic categorization (no LLM required)
- MIT license, tests, CI