Contact

Join the waitlist

Tell us about the AI applications and agents you want to test. Reinstator is available by waitlist, and we will be in touch by email.

FAQ

Questions

Something not covered here? Ask in the note field when you join the waitlist.

Back to the sign-up form

What can Reinstator test?

Two things. Reinstator red-teams AI applications and tool-using agents across tool routes, indirect channels such as retrieved documents and tool outputs, and persistent state such as memory, policy, workflows and skills. It also runs an AI-driven pentest across a whole application, over browser journeys, API sequences and AI surfaces.

Where can Reinstator run?

Anywhere from development to production. Reinstator reaches only the endpoints and paths your plan authorizes, and stays inside the rate and time limits you set. In agent tests it never executes your tools; your gateway runs them inside the isolated test sessions it provides, and every attempt starts a fresh one.

What is fresh-session replay?

When an attack succeeds, the exact winning messages run again in new sessions. The result becomes a finding only if every replay reproduces it.

Where is evidence stored, and who can read it?

In Reinstator's encrypted evidence store, bound to your tenant and job. Plans, transcripts, findings and reviews are encrypted; IDs, status and timestamps stay in plaintext so jobs can be queried. Only reviewers can read raw transcripts, and every read is logged.

How is spend capped?

Every call's declared cost ceiling is reserved against the run's budget before the call is sent. A run stops rather than send a call that would exceed its request or dollar budget, and it stops when its time budget runs out.

Does Reinstator block attacks at runtime?

No. Reinstator tests the boundaries your agent enforces. It does not sit in front of your agent or block traffic.

How do I get access?

Reinstator is available by waitlist. Join with your work email and tell us what you want to test. We will be in touch by email.