Your pod, with code in it.

A pod is a small piece of storage you own. With Mind Codespaces, four kinds of things go inside it — and nothing else needs to leave.

The flow has three beats. There's some plumbing — a small bridge translates Git Smart HTTP into pod writes — but you only ever talk to two things: your terminal and your pod.

1. 01

   You push

   git push, like any other remote.

2. 02

   The bridge translates

   Smart HTTP in, pod writes out. No project state stored here.

3. 03

   Your pod receives

   Code, metadata, and the live site land under your WebID.

The coder agent runs against the same pod the user owns, with the permissions the user granted. It reads what you let it read, writes where you let it write, and the conversation history — the thing AI tools usually keep in their own database — lands back in the pod alongside the code. If you revoke access, the agent stops. If you move pods, the memory comes with you.

• Readsissue body, recent commits, files the prompt names — nothing else until you say so.
• Writesa branch in your repo, a pull request, screenshots and comments in your pod.
• Remembersthe conversation, but the memory lives in the pod, not in an opaque vendor database.
• Stopsthe moment you revoke its permission — the bridge is a delegate, not an owner.

The bridge can be replaced and your work survives, because nothing irreplaceable lives in the bridge.