memsearch

Why we included this project

Every team that leans on AI coding agents has hit the moment where the assistant forgets what it decided a week ago. memsearch gives Claude Code, Codex, and similar tools a memory that survives between sessions: what an agent reads and concludes is stored as Markdown, indexed with Milvus embeddings, and pulled back through hybrid or semantic search when that context matters again. The project also lets you turn recurring workflows into installable skills, so an agent gradually gets better at your particular codebase. You can use it as a standalone CLI, as a Python library, or as a plugin for the coding CLIs you already run, which keeps adoption simple.