# OzBrain

> OzBrain is a hosted, structured knowledge base ("brain") that AI agents read and write through one MCP connector. Claude, ChatGPT, Claude Code, and Cursor share the same linked articles, found through a routing index, written through staged writes with contradiction checks and per-author provenance. Brains can be shared with other accounts, read or read and write, so a team's agents work from one current version of the truth. Full markdown export, anytime.

## Product

- [OzBrain product description](https://ozbrain.com/docs): current capabilities and limits, with an agent-readable markdown version at https://ozbrain.com/docs.md
- [About OzBrain](https://ozbrain.com/about): founder, company, and product context.
- [OzBrain for Teams](https://ozbrain.com/use-cases/teams): one shared brain every teammate's AI reads. Writes follow the assigned role.
- [Security](https://ozbrain.com/security): tenant isolation, audit log, revoke, encryption at rest.
- [OzBrain for Claude](https://ozbrain.com/for/claude): connector setup in Claude.
- [OzBrain for ChatGPT](https://ozbrain.com/for/chatgpt): connector setup in ChatGPT.
- [OzBrain for Claude Code](https://ozbrain.com/for/claude-code): one-command setup for Claude Code.
- [OzBrain for Cursor](https://ozbrain.com/for/cursor): settings entry for Cursor.
- [OzBrain for OpenClaw](https://ozbrain.com/for/openclaw): MCP setup across OpenClaw channels.
- [OzBrain for Hermes Agent](https://ozbrain.com/for/hermes): remote MCP login for Hermes.
- [OzBrain for Gemini](https://ozbrain.com/for/gemini): Gemini Spark custom MCP app setup.
- [Claude connector docs](https://ozbrain.com/docs/claude-connector): setup prompts and tool walkthrough.

## Agent guides

Hand one of these URLs to an agent. Each serves markdown, not a web page, so a server-side fetch reads the whole thing on the first try. A `.md` twin of each path serves the identical document.

- [Connect OzBrain](https://ozbrain.com/docs/connect): step-by-step connector setup for Claude, ChatGPT, Cursor, and any other MCP client, then sign-in and a confirmed connection.
- [Start using OzBrain](https://ozbrain.com/docs/start): save the first real articles from the person's own context, show them what landed, then ask consent for the ongoing save and check habits.
- [Evaluate OzBrain](https://ozbrain.com/docs/learn): judge from the person's own work whether OzBrain would help them, in under 150 words.

## Starter templates

- [Obsidian vault import](https://ozbrain.com/templates/obsidian): send the template link to a connected desktop agent; installer markdown at https://ozbrain.com/templates/obsidian.md

## Comparisons

- [OzBrain vs ChatGPT Memory](https://ozbrain.com/compare/ozbrain-vs-chatgpt-memory): shared brain vs a digest one product keeps.
- [OzBrain vs Claude Projects](https://ozbrain.com/compare/ozbrain-vs-claude-projects): knowledge under every agent vs uploads for one workstream.
- [OzBrain vs ChatGPT Projects](https://ozbrain.com/compare/ozbrain-vs-chatgpt-projects): knowledge under every agent vs one workspace box.
- [OzBrain vs Notion](https://ozbrain.com/compare/ozbrain-vs-notion): agent brain vs where the team writes.
- [OzBrain vs Obsidian](https://ozbrain.com/compare/ozbrain-vs-obsidian): shared agent brain vs a local vault.
- [OzBrain vs CLAUDE.md files](https://ozbrain.com/compare/ozbrain-vs-claude-md): one brain vs a copy per repo.
- [OzBrain vs a DIY LLM wiki](https://ozbrain.com/compare/ozbrain-vs-llm-wiki): hosted maintenance vs being the librarian.
- [Mem0 vs Supermemory vs OzBrain](https://ozbrain.com/compare/mem0-vs-supermemory-vs-ozbrain): memory APIs for builders vs a brain you connect.

## Policy

Crawler access is controlled by robots.txt on this host. Search-index, user-fetch, and training bots are allowed. llms.txt is agent-readable orientation only; it does not grant or deny crawl access.