agentmemory

Why we included this project

Most coding agents forget everything between sessions, which is why you end up re-explaining your architecture and re-discovering the same bugs. agentmemory fixes that with a local memory server that watches what your agent does, compresses it into searchable memory, and feeds the right context back when a new session starts. It works with Claude Code, Copilot, Cursor, Codex, and any other agent that speaks MCP, hooks, or REST, and one server shares memories across all of them, so switching tools between projects doesn't mean starting from scratch. The project ships benchmarks like LongMemEval-S and token-cost comparisons to give you a concrete sense of retrieval quality, and local embeddings mean recall works without an API key. If you want your agents to actually remember project decisions across sessions, this is a practical way to get there.