Mori

mori (森)

A shared memory layer for AI coding agents —
one that compounds.

Get started → View on GitHub ↗

The problem

One codebase. Many agents. No shared memory. Until now.

Your team shares a repo, standards, and a CI pipeline. But each agent session starts fresh — no memory of what the others decided, changed, or broke. Mori fixes that.

One forest, many agents

How it works

01

Capture

Claude Code lifecycle hooks send session events to Mori automatically. Every tool call, decision, error — captured without changing how you work.

02

Dream

While you sleep, the dream pipeline distils events into durable memories. Signal, not transcripts. What matters, not what happened.

03

Brief

Every session starts with /brief — shared memories, team standards, cross-instance decisions. From turn one, every agent knows what the others know.

The dream pipeline: capture, distil, brief

No instrumentation required

Every other memory tool requires you to instrument your app — call add() after responses, call search() before prompts.

Mori requires nothing. Lifecycle hooks capture your sessions automatically. You just work.

Works with your agent

Claude Code
Cursor
Antigravity
Cline

Get running in minutes

git clone https://github.com/fjwood69/mori.git
cd mori
cp deploy/homelab/.env.example deploy/homelab/.env
docker compose -f deploy/homelab/docker-compose.yml up -d

# Verify
curl http://localhost:8968/health
{"status":"ok"}
Architecture: one server, many agents

For teams

One Mori, many agents. Every session benefits from what every other session learned.

Point every instance at the same server. That's it.

The forest remembers. Now it whispers too — tasks and decisions passed directly between agents, without a shared session.

The forest whispers — inter-agent messaging