Claude-Mem
Lets Claude Code remember earlier sessions so you can pick up where you left off
by Alex Newman
ai-memory
by Fabio AkitaCLI toolWrite code
Gives your coding agents a shared memory so the next one picks up where you stopped
The problem it solves
You hit a usage limit in one coding agent, open another, and have to explain the whole project and what already failed all over again.
For DevelopersPeople who use several coding agents
- Before
- Each coding agent keeps its own notes on one machine, so a new session starts without the old context.
- After
- The next agent gets a handoff that says where you stopped, what failed and what is still open.
- Stars
- ★ 15k
- License
- MIT
- Cost
- Free
- Platforms
- Mac · Win · Linux
- Level
- Intermediate
Overview
ai-memory is a program that runs on your computer and records what your coding agents do, after sanitizing it, as plain markdown pages in a git-backed wiki. When you quit one agent and open another, such as Claude Code and then Codex, the next one gets a handoff that says where you left off, what failed and what is still open. It works with more than twenty agents and needs no model API key for its default features. It suits people who switch between coding agents.
The memory is ordinary markdown files in git, so you can read, search and edit it by hand.
Try these first
Start with the first one: it is a small task you can check easily.
- What do you remember about this project?First try · read-only
- Write a handoff for the next agent: where I stopped, what failed and what is still open.writes memory
- Search memory for how we decided to structure the database.read-only
- Continue from the last handoff in this folder.reads memory
Example prompts written by Agent Flows Hub.
Features
From the creator's documentation.
| Feature | What it does | Runs |
|---|---|---|
| Handoffs | Passes where you stopped, what failed and what is open to the next agent | Your computer |
| Automatic capture | Hooks record sanitized prompts and tool calls as you work | Your computer |
| Markdown wiki | Memory is plain pages in git that you can search and edit | Your computer |
| No model key needed | Capture, search and handoffs work without an LLM API key | Your computer |
| Agent setup | Wires MCP, hooks and skills into the agents it finds, after asking | Your agent settings |
| Shared server | Optional: one server for several machines or a team | Your server |
Good to know
You use more than one coding agent, or hit a usage limit in one and switch to another. ai-memory keeps a shared record of the project and hands the next agent a summary of where you stopped, so you do not explain it again.
Limits
- Native Windows support is marked experimental; macOS, Linux and Windows through WSL2 are listed as supported.
- Setup changes the configuration of the agents it finds: it shows its plan first and asks before applying it.
Notes by Agent Flows Hub (original project name: ai-memory), written from the creator's documentation, not from running it ourselves. ai-memory README · ai-memory support matrix
Stuck setting it up?
Tell us what happened. We take a look first; if it needs paid work, you get an estimate before anything starts.
More to write code
Other cli tools
- ccusage Shows how many tokens your coding agents used and what that would cost, day by day
- Claude Code Templates Installs ready-made skills, agents, commands and MCP setups into Claude Code in one step
- Claude Code Usage Monitor Shows your Claude Code usage live in the terminal and warns you before you hit a limit
- Claude Token Optimizer Shrinks the notes Claude Code rereads at every session start, so each uses fewer tokens
- CodeBurn Shows what your AI coding tools cost, split by project, model and task
- Dev Browser Lets your coding agent drive a real Chrome browser to open pages, click and check results