knowledge-base-server: a shared memory store for AI agents
An open-source, self-hosted knowledge base that ingests notes, code, and documents and exposes them via MCP and REST so Claude Code, Codex, ChatGPT, and other clients share retrieval context.
Your AI agents keep rediscovering the same context. This repo gives them a shared memory. knowledge-base-server is a self-hosted persistent memory system for builders who use AI agents across projects. It helps you turn notes, code, documents, and session findings into retrievable context by ingesting them into a shared knowledge base exposed through MCP and REST APIs. Key features: • Shared agent context – Claude Code, Codex CLI, Gemini CLI, ChatGPT, and other MCP clients can read from and write to the same knowledge base • Hybrid retrieval – FTS5/BM25 search and local semantic search help agents find relevant material • Broad ingestion – supports Obsidian sync plus code, PDFs, text, YouTube transcripts, X bookmarks, web articles, and terminal sessions • Knowledge pipeline – tools classify captures, promote raw sources into structured artifacts, and synthesize patterns across notes • Token-aware briefings – `kb_context` returns summaries and metadata before full documents are loaded It’s open-source (MIT license). Link in the reply 👇