ai-memory is an open-source tool that gives AI coding agents a shared long-term memory of a project, so a new session or a different agent can carry on where the last one stopped.
In plain words
Coding agents usually keep their own notes, which stay with one agent on one machine. ai-memory keeps context outside the agent: hooks record your prompts and some tool activity, and at the end of a session those notes become Markdown pages in a project wiki kept in git. The next session, in the same agent or another, starts with a handoff note listing where you stopped, which attempts failed and which questions remain open. Basic features need no language-model API key, and it works with 35 coding agents.
A quick example
Imagine you run out of usage in one coding agent halfway through fixing a bug. You open a different agent in the same project, and it starts with a note about what you tried, what broke and what is left, so you don't have to explain everything again.
Where you'll meet it
- Switching between coding agents in the middle of a task.
- Sharing one project memory across a team's agents.
Related
Sources
- Docsai-memory: long-term memory for AI coding agents, ai-memory · read 22 Oct 2026
- Docsakitaonrails/ai-memory, ai-memory (GitHub) · read 18 Oct 2026
Checked by WhatIsAI against the sources above · 22 Oct 2026. A full five-minute explainer with a diagram may follow. Spotted an error? Report it.