capability-diagnostic

Agent failure diagnosis involves analyzing tool availability, context, model capabilities, task specification clarity, and infrastructure. This Skill applies a structured decision tree to pinpoint the root cause and recommends actionable fixes for improving agent performance.

The best way to troubleshoot agent failures is by systematically matching common failure patterns using a decision tree. This approach determines whether the issue stems from missing tools, insufficient context, model limitations, or infrastructure problems, and then recommends specific fixes.

You can identify whether model capabilities or task specifications caused the agent failure by running a diagnostic analysis. The Skill systematically evaluates both the model tier's suitability for the task complexity and the clarity of the provided prompt to isolate the root cause.

You need root cause analysis for agent failures when an agent repeatedly fails to complete complex tasks. This diagnostic process is essential for troubleshooting and improving agent performance by determining if the failure is due to missing libraries, ambiguous prompts, or infrastructure issues.

The limitation of troubleshooting agent failures with a decision tree is that it relies on matching common patterns within the provided context. If the failure involves unprecedented edge cases outside the analyzed infrastructure or model capabilities, the diagnostic may require additional manual investigation.