Get free $30 in credits on usBrain is now live for personal and enterprise use. Join leading teams transforming with AI.Join the waitlist

The brain behind your Cursor sessions

Cursor is fast because it lives in your code. It is forgetful because your knowledge lives everywhere else. Connect the two.

Last updated

AIVM Brain gives Cursor governed, persistent memory over MCP. After connection through your workspace, Cursor can recall your decisions, docs, and project facts, and capture new ones as you work. Permission-aware for teams, audited end to end, free to start.

How do you connect Cursor?

Connect Cursor using the current setup instructions inside your AIVM Brain workspace. Public access currently starts with the waitlist.

  1. Get workspace access. Join the waitlist with your work email and confirm your email. If you already have access, log in to your workspace.

  2. Follow the in-app setup for Cursor. Use the current connection instructions provided in your workspace for Cursor. Follow the access scope assigned to your agent.

  3. Check access and retrieval. Ask Cursor a question about an approved source and check its citation. Also test a restricted source to confirm that the agent cannot retrieve content outside its permissions.

Rules files stop at the repo boundary

Cursor's rules are per-project by design. That is correct for coding conventions and exactly wrong for knowledge. The reason you chose Postgres over Dynamo, the auth quirk that bit you in March, the vendor API's undocumented rate limit: none of that belongs in a rules file, and all of it is what you re-explain to your editor every week.

The brain sits outside the repo, so your Cursor sessions share one memory across every project and every machine. Ask in the editor, get the answer with its provenance, keep moving.

Your whole team's context, scoped per person

On a team, Cursor plus a shared brain means the frontend dev's editor knows what the platform team decided last week, without anyone writing a wiki page about it. Retrieval is permission-aware: each member's key sees what their role allows and nothing more, with field-level redaction for the sharp edges (a doc can be recallable while its salary column is not). Every recall is written to a tamper-evident ledger, which is what makes a shared brain survivable in a security review.

Questions, answered

Does this conflict with Cursor's own memory features?

No. Anything Cursor keeps locally stays as is. The brain adds a governed, shared memory over MCP that also works from your other agents and tools.

Do I have to change how I work?

No. Ask Cursor things as usual; when the brain has the answer, the MCP tools surface it. Capturing a fact is one instruction when something is worth keeping.

Can I keep some knowledge personal, not shared?

Yes. Solo content stays yours. On a team, sharing follows roles and grants; nothing becomes team-visible by accident, and the ledger records every access.

Does it slow the editor down?

Recall is a tool call Cursor makes when it needs context, not a proxy in front of your model. Editor performance is untouched.

What does it cost?

Free to start. Team and enterprise tiers are public at brain.aivm.io/pricing.

Give Cursor a brain it can be held to.