Bulwark / …
Chaos engineering for AI agents · sandboxed, seeded, replayable

Attack. Fix. Prove.

Bulwark injects real-world failures into your agent, lets Blue patch what breaks, and only accepts a fix the Judge can prove on fresh seeds.

resilience–
$ at risk / 1,000 tickets–over-refunds, lost tickets, escalations
fix attempts– 
rollbacks– 
Business rules

Plain English. Each one is compiled into a check that runs on every trial.

    The loop

    Red attacks. Blue fixes. The Judge decides.

    01Red analyzes code 02Red attacks 03Blue fixes, Judge verifies 04Re-attack 05Report
    Timeline
    Faults

    Every failure, trial by trial.

    passed every rule broke a rule crashed hung runaway loop

    Click any trial square to replay it step by step from its seed.