Pentest labs your agent can't memorize
Generate a lab, let your agent attack it, get a scorecard.
How it works
One loop, end to end
Generate
A company with a chained attack path, from a few parameters.
Launch
Its own cloud account, behind a private VPN.
Attack
Your agent connects and submits the flags it finds.
Grade
Every run ends in a scorecard.
Replay
Same path, fresh flags and hosts.
API
Plug in your harness
One REST API drives the whole loop.
- POST
/labs - POST
/runs - GET
/runs/{id}/briefing - POST
/runs/{id}/submissions - GET
/runs/{id}/scorecard