brethof-brain

Every AI agent forgets the moment a session ends, and you become its memory: re-explaining the project, the conventions, what was already tried. brethof-brain is the memory instead.

What your agent gets, without asking:

• At session start: your rules, what each project is, where the last sessions stopped, and the plan's next lines.
• With every prompt, before it answers: the records that bear on it and the rule it touches. No tool call.
• Everything ever said stays searchable, and every decision is kept as a dated line (decided, tried, rejected, and why) so nobody re-proposes the idea you already killed.
• One memory across 20 proven platforms: Claude Code, Codex, Gemini CLI, GitHub Copilot CLI, Hermes Agent, OpenClaw and more.

Privacy is the design, not a setting. The memory lives on your own machine, or in our cloud, encrypted in Germany under a passphrase only you hold. The processing runs in a secure compute instance in Zurich: it keeps a piece only long enough to understand it and stores none of it. The thinking runs on a model under zero data retention.

Install: tell your agent "install brethof-brain". The client is source-available.

Free plan: 500 messages a month, no card. Paid plans from $9.99/month at founder price, the same price everywhere.