agent-failure

Observability and enforcement for AI agent harnesses. Capture every run and runtime reliability with policy enforcement.

Observability and guardrails for your ai agents. AgentTrail Guard traces your agent's execution and automatically compiles repeat failures into permanent guardrails.

Agent Failure Series #10 — EscapeHatch: Interactive simulator of AI agent sandbox escape. Based on the live OpenAI/Hugging Face incident, July 2026.

BLEEDTHROUGH — Agent Failure Series #20. Multi-tenant context isolation failure: shared AI support agent bleeds one user's data into another's session.

Agent Failure Series #13 — ContextDrift: interactive simulator of attention-weight decay in long-context agents

Agent Failure Series #23 — Safety rules evicted as context fills. An interactive demo showing how AI agents silently become dangerous when their context window compresses safety constraints.

TOOLROT — Agent Failure Series #21: Tool selection collapses when descriptions are semantically similar. Interactive demo.

Interactive simulator showing how AI agents fail silently — HTTP 200, valid JSON, zero error signals, real task never executed.

GhostApproval — Agent Failure Series #17. Interactive simulator: 4 failure modes where AI agents bypass human approval gates. HTTP 202 confusion, timeout default-allow, swallowed exceptions, cached approval bypass chain.

Agent Failure Series #22 — ECHOBURN: The idempotency blindness demo. Agent reports ✅ $99.99 while Stripe shows -$399.96.