CEREBRO.md

Your AI assistant's memory — a persistent knowledge system that works across every coding agent.

Every AI session starts from scratch. Cerebro fixes that. It records your sessions, captures your knowledge, and loads your context automatically — so your AI picks up right where you left off.

Get Started View on GitHub

$ /cerebro start

Hey! Welcome back.

Health Check       ✓ All systems go
Last Session       Yesterday — shipped MCP server for Claude chat
Streak             7 days
Pinned             Write tests for research mode
Reminder (due)     Review PR #42

Recent Activity:
  • 2026-06-25 — MCP server registered in 6 platforms
  • 2026-06-24 — Cerebro skill deployed to Cursor + Codex CLI
  • 2026-06-23 — Shipped v0.2.0 — 57 commands, research mode

What would you like to work on today?

Why Cerebro?

AI coding assistants have no memory between sessions. Every conversation starts from zero. Cerebro fixes that.

Persistent memory

Every session builds on the last. Notes, decisions, TILs, and activity history accumulate in plain markdown — readable forever, no lock-in.

Works everywhere

Claude Code, Claude Desktop, Cursor, Windsurf, Cline, Codex CLI, Gemini CLI, Google Antigravity. One install, one /cerebro command, everywhere you work — with MCP giving full read/write access in chat UIs.

Your data, your machine

Nothing leaves your computer unless you choose cloud storage. Opt in to exactly what you want tracked. Export or delete everything with one command.

Works with every AI coding agent

One install. One command. All your tools.

Claude Codeauto-logskill

Claude DesktopMCP

CursorskillMCP

Codex CLIskillMCP

AntigravityMCP

WindsurfMCP

ClineMCP

Gemini CLI

auto-log — sessions recorded via hooksskill — Cerebro context loads automaticallyMCP — full read/write access in chat

Everything you need to never lose context again

Session Recording

Automatically logs every prompt and response. Know exactly what you worked on, when, and for how long.

Activity History

An append-only journal across all sessions. Never lose track of what you did last Tuesday.

Knowledge Capture

Notes, decisions, code snippets, TILs, bookmarks — all searchable, taggable, and linked.

Research Mode

Ingest articles, papers, or repos into a self-organizing wiki. Every source becomes linked concept pages — and every query files its answer back in, so the wiki compounds over time.

Smart Start

Every session begins with context. Recent activity, reminders, goals, and pinned items — all surfaced automatically.

Privacy Controls

Anonymous mode, private sessions, granular consent, full data export, one-click delete.

Token Tracking

Know exactly how many tokens each session uses. Track costs across projects.

Daily Dashboard

/cerebro today — everything from today in one view: notes, sessions, reminders, goals.

Handoffs & Standups

Generate structured handoff docs to resume in a new session, or export a standup update from your recent activity — copy-paste ready.

Plain Markdown

Sessions, notes, and knowledge live in ordinary .md files — readable, diffable, and yours forever.

MCP Server

A built-in MCP server gives Claude Desktop, Cursor, Codex CLI, Antigravity, Windsurf, and Cline full read/write access to your Cerebro data — even in chat interfaces without slash commands.

How it works

Install

Clone the repo and run bash install.sh. Takes 30 seconds. Detects your AI tools automatically.

Onboard

Type /cerebro in any supported AI assistant. Answer a few questions about your preferences and privacy settings.

Work

Start every session with /cerebro start. Capture notes, decisions, TILs mid-session. End with /cerebro end.

Remember

Next session, your AI assistant picks up right where you left off — context, goals, reminders, and history all loaded.

57 commands for your entire workflow

All commands invoked as /cerebro <command>. Expand a category to see commands.

Session

/cerebro start/cerebro end/cerebro private/cerebro log

Capture

/cerebro note/cerebro til/cerebro snippet/cerebro decision/cerebro inbox/cerebro bookmark/cerebro pin

Focus

/cerebro goal/cerebro blocked/cerebro checkin/cerebro workspace/cerebro context

/cerebro search/cerebro find/cerebro query/cerebro tag

Knowledge

/cerebro new-project/cerebro new-kb/cerebro research/cerebro memory/cerebro map/cerebro graph/cerebro links

Review

/cerebro today/cerebro digest/cerebro review/cerebro changelog/cerebro flashback/cerebro stats/cerebro streaks/cerebro patterns

Collaboration

/cerebro team/cerebro join/cerebro share/cerebro handoff/cerebro standup/cerebro mention/cerebro delegate/cerebro push/cerebro pull

Automation

/cerebro template/cerebro alias/cerebro watch

System

/cerebro status/cerebro profile/cerebro consent/cerebro privacy/cerebro clean/cerebro update/cerebro setup/cerebro uninstall/cerebro export

Organization

/cerebro archive/cerebro fav/cerebro dedup

See it in action

End a session and Cerebro summarizes what you did, updates your history, and backs up your memory.

$ /cerebro end

Session Summary:
  • Built Cerebro v0.2.0 — 32 new commands, research mode
  • Added Claude skill file (cerebro-skill.md)
  • Merged remote platform support (Codex, Gemini, Antigravity)
  • Pushed to github.com/rcktshp/cerebro.md

Goal Check: Did you make progress on "Ship v0.2.0"? → marking complete ✓

Activity history updated  →  activity-history.md
Memory backed up          →  memory-backup.md
Time: 2h 14m  |  Turns: 47  |  Session: 2026-06-25-094312.md

Privacy first, always

You own your data. Cerebro never phones home.

100% local by default

All data stays on your machine. Nothing is sent anywhere unless you choose cloud storage.

Opt-in tracking

Session recording, activity logging, token tracking — every feature is opt-in during onboarding.

Private mode

/cerebro private — start a session with zero logging. Nothing recorded, nothing stored.

Anonymous mode

Strips your name from all content so files contain no personal identifiers.

Full export

/cerebro export — all your data, portable, in a folder or zip. Yours to take anywhere.

Delete everything

/cerebro privacy delete-all — removes every file Cerebro has ever created.

Tiny footprint. No surprises.

No databases. No background processes. No network calls. Just markdown files and JSON.

Location	Contents	Size
~/.cerebro/	Config, hooks, MCP server, session map	~20 KB
~/Cerebro/	Sessions, notes, history, knowledge	grows with use
platform configs	MCP entries merged into each app's config	~1 KB each

Python 3 only (stdlib, no pip)macOS · Linux · WindowsNo network calls at runtime

Pricing

Open Source

Freeforever

Full personal use. All commands, all platforms, all features.

✓57 slash commands
✓All 8 platforms
✓MCP server for Claude chat & more
✓Session recording + activity history
✓Research mode + knowledge base
✓Privacy controls
✓Self-hosted

Install now

Team

Coming soon

Share decisions, handoffs, and standups across a team via git.

✓Everything in Open Source
✓Shared storage layer
✓Team mentions + delegate
✓Push / pull sync
✓Standup aggregation

Join waitlist

Enterprise

Custom deployment, SSO, audit trails, and dedicated support.

✓Everything in Team
✓SSO / SAML
✓Audit logs
✓Custom integrations
✓Dedicated support

Get in touch

Get started in 30 seconds

Then open your AI assistant and type /cerebro to begin onboarding.

git clone https://github.com/rcktshp/cerebro.md.git
cd cerebro.md
bash install.sh

Requires Python 3 · macOS · Linux · Windows