Tool selection errors happen when an AI agent chooses an inappropriate tool, often due to incorrect framing in the system prompt or improper tool availability.

Six common AI agent failure types include context degradation, specification drift, sycophantic confirmation, tool selection errors, cascading failure, and silent failure.

Prompt engineering specifying what you want a system to output was adequate when AI systems were stateless single-turn tools under direct supervision, but is structurally inadequate for long-running autonomous agents.