kormem

Local long-term memory for AI agents. kormem indexes everything you say to assistants and every file, folder or URL you add, files it into an Obsidian-compatible vault of linked notes, and answers questions with a small recall agent that returns dated findings and their sources.

The commands follow QMD: collection add, context add, search, vsearch, query, get, multi-get and ls work the same way.

pip install kormem            # or: uvx kormem …, npm install -g kormem, bunx kormem …
kormem init                   # key file, engine, background services, Claude Code plugin

kormem collection add ~/notes --name notes
kormem context add kormem://notes "Personal notes and ideas"
kormem search "project timeline"            # keyword search
kormem vsearch "how to deploy"              # semantic search
kormem query "what did we decide about X?"  # recall agent (best quality)
kormem get "#abc123"                        # open a source by docid
kormem remember "I prefer pnpm over npm"
kormem mcp                                  # MCP server for Claude Code, Claude Desktop, Cursor

kormem needs one key: put GEMINI_API_KEY in ~/.kormem/.env, which kormem init creates. It runs embeddings, extraction and the recall agent. Without it, kormem still indexes and searches locally. kormem doctor checks the setup.

Benchmark

LongMemEval-S, all 500 questions, scored by the official GPT-4o judge. Each question gets one answer from a curated context of about 19k tokens.

Other systems' numbers are self-reported. Setups and caveats: full report.

Full documentation, diagrams and the Claude Code plugin: https://github.com/kushalsinha/kormem