A Rust-based CLI and MCP tool for sharing memory and task handoffs across coding agents
ai-memory, written in Rust by Fabio Akita, is a long-term memory layer for coding agents. Memory travels across agents, projects, and machines. More than 20 agents—including Claude Code, Codex, Cursor, Gemini CLI, OpenCode, and Kiro—share the same memory. If you stop one agent mid-task and start another, the new agent picks up with a handoff note. It says where you left off, what failed, and what’s still open. Handoffs use a typed protocol: each handoff has an owner and can be claimed only once.
The source of truth is a Git-versioned wiki of plain Markdown files. A SQLite FTS5 index can always be rebuilt from those files. By default, capture, search, and handoff make no LLM calls and require no API key. Hooks sanitize prompts and tool calls at a privacy boundary before logging them. Multi-user authentication and audit logging aren’t reserved for a paid tier.
Regolo.ai published a test on October 28, 2026. In the test, the prompt tokens needed to get up to speed on a repository fell from 26,185 to 5,453 (79%). In a 90-turn handoff from Claude Code to OpenCode, they dropped from 52,262 to 2,485. The benchmark used optional LLM consolidation and the testing provider’s own endpoint. Don’t expect savings on small, one-off tasks. The project has 14.8k stars and more than 100 contributors. The latest release, v3.3.1, came out on November 1, 2026.
People who switch between agents such as Claude Code, Codex, and OpenCode during the day. Developers working across multiple machines or looking to share agent memory with a team.
You need to run your own server. The Docker quick start has no authentication, so use it only on a network you trust. Hooks log prompts and tool calls; check the privacy settings to see what’s stored. The independent benchmark used LLM consolidation; savings from the default zero-LLM path haven’t been measured separately.
curl -fsSL https://github.com/akitaonrails/ai-memory/releases/latest/download/install.sh | sh ai-memory setup claude plugin marketplace add akitaonrails/ai-memory claude plugin install ai-memory@ai-memory