Apply & Reference

Failure Modes

How production AI systems fail — and how to detect and prevent it.

Hallucination

Ungrounded generation treated as fact.

Coming Soon

Prompt Injection

Untrusted content hijacks instructions or tools.

Coming Soon

Context Poisoning

Hostile material enters the assembled context window.

Coming Soon

RAG Poisoning

Indexed content is planted to steer answers or actions.

Coming Soon

Memory Poisoning

Persisted memory is corrupted and reused later.

Coming Soon

Goal Drift

The agent silently optimizes a different objective.

Coming Soon

Infinite Agent Loops

The loop never terminates or retries without bound.

Coming Soon

Tool Misuse

A valid tool is called with the wrong intent or arguments.

Coming Soon

Excessive Permissions

The agent holds more authority than the task requires.

Coming Soon

Confused Deputy

The agent uses a user's authority against that user.

Coming Soon

Capability Chaining

Tools are composed into an unauthorized higher privilege.

Coming Soon

Credential Leakage

Secrets enter prompts, logs, or model context.

Coming Soon

Duplicate Actions

Retries cause double charges, tickets, or writes.

Coming Soon

Retry Storms

Failure amplification across models, tools, and queues.

Coming Soon

Stale Authorization

A grant outlives the task, risk, or user session.

Coming Soon

Cross-Tenant Leakage

One tenant's data appears in another's context.

Coming Soon

Provider Failure

Model or tool outages without a designed fallback.

Coming Soon

Silent Retrieval Failure

Empty or wrong retrieval is treated as a valid answer.

Coming Soon