Open Second Brain
Memory layer for AI agents that lives in an Obsidian vault: o2b CLI, MCP servers and hooks for Claude Code, Codex, Cursor and other agents.
590
GitHub Stars
Obsidian
Vault Storage
MIT
License
Overview
Open Second Brain is a memory layer for AI agents that lives inside an Obsidian vault. It installs an o2b CLI, MCP servers, lifecycle hooks and client plugins, so agents such as Claude Code, Codex, ChatGPT desktop, Cursor, Gemini CLI, Copilot CLI and Aider can read and write the same markdown memory under a Brain/ folder. A "dream" pass learns your preferences, every change is attributed in a hash-chained log, and brain changes can be rolled back from snapshots.
The Verdict
Who Should Use Open Second Brain?
Best For
- Obsidian users who want their agents to share one memory
- People who use several agents and MCP hosts
- Anyone who wants reversible, auditable agent memory
Not Ideal For
- Users who want a one-click install
- Teams needing a hosted multi-user service
What's Great
- Local-first markdown you can read, edit and version
- Reversible brain changes with
o2b brain rollback - Hash-chained, attributed write log
- Works with any MCP host
Watch Out For
- Needs Bun 1.1+ and several install steps
- Semantic search and some features are off by default
- Young project with a small community
Pricing
View all features & details
Key Features
- Preference learning via a "dream" pass
- Snapshots and rollback
- Keyword search plus optional semantic search
- Session recall from agent transcripts
- Rebuildable SQLite index
Use Cases
- Personal knowledge base your agents maintain
- Cross-agent memory in one vault
How It Compares
| Feature | Open Second Brain | ai-memory | Claude Second Brain |
|---|---|---|---|
| Storage | Obsidian vault | Git markdown wiki | Obsidian vault on GitHub |
| Agents | Any MCP host | 20+ agents | Claude Code first |
| Rollback | Yes, snapshots | Git history | Git history |
| License | MIT | MIT | MIT |
Loading reviews...
Is this your tool? Add a badge