What problem does it solve? Teams often ship workflows, APIs, and AI-provider integrations without proving how they behave under timeouts, corrupted state, rate limits, or policy failures, leaving release-critical behavior unverified. ## Core Features & Use Cases - Failure Classification: Categorize each failure as fail-closed, degrade-with-recovery, or bounded-retry based on security, compliance, and business impact. - Deterministic Scenario Design: Define controlled fault-injection scenarios using fake providers, fixture corruption, injected clocks, and Playwright route stubs instead of random production chaos. - Evidence Reporting: Produce a structured evidence report mapping scenarios to acceptance criteria, recovery paths, audit events, and unresolved resilience gaps. - Use Case: Before releasing a feature that depends on an external LLM provider, define scenarios for provider timeout, budget exhaustion, and offline mode, then capture proof that the system degrades safely and logs the right audit events. ## Quick Start Ask the assistant to design chaos and resilience test scenarios for your workflow, covering provider timeouts, corrupted state, and rate-limit exhaustion with an evidence report.