nbonamy/witsy

Supports multimodal AI across 20+ providers (vision, image/video generation, text-to-speech, and speech-to-text) with a BYOK model, plus document RAG via local embeddings and Anthropic Computer Use capabilities. Functions as a universal MCP client enabling any LLM to run MCP servers through stdio transport. Includes native OS integration for "Prompt Anywhere" and AI commands accessible via global shortcuts in any application.

1,905 stars and 20 monthly downloads. Actively maintained with 32 commits in the last 30 days. Available on npm.