Launching today
Memtru
Test AI agent memory for leaks and deletion failures
1 follower
Test AI agent memory for leaks and deletion failures
1 follower
Memtru helps developers test AI agent memory for cross-user leaks, stale information and deleted data reappearing after background work. Run the Python CLI locally, reproduce failures, and compare reports before and after a fix.
Free
Launch Team
CloudflareYour next app belongs on Cloudflare Workers. Deploy for free
Promoted
Hi Product Hunt!
Memtru tests the memory behind AI agents: can one user retrieve another user's information, does an update leave stale facts active, and can a background job bring deleted information back?
The Python CLI runs locally and produces JSON and HTML reports with evidence you can reproduce. Connect your own system through an adapter, run tests in an authorized sandbox with synthetic data, then rerun after a fix. The hosted beta dashboard lets you review explicitly uploaded reports.
The public integration demo shows the same deletion test failing against an intentionally vulnerable application and passing after a fix. It uses a synthetic application and a separate background worker, not a customer deployment.
A passing run describes the scenarios and retrieval paths tested; it is not a guarantee of absolute security or physical data erasure.
Start with the demo at https://memtru.com/demo/integration or the CLI: pip install memtru.
Feedback that would help most: what memory stack are you using, and which deletion or isolation failure is hardest for your team to reproduce?