investigate

To find the root cause of a software bug, this skill collects evidence from code, logs, and test artifacts by dispatching parallel investigators to trace symptoms to their true origin. It produces a structured evidence summary mapping findings to specific file paths and lines.

The best way to debug an unknown failure is evidence-based investigation, which dispatches multiple parallel agents to analyze error paths and data flows concurrently. This coordinates concurrency, behavioral, and data specialists to pinpoint the true cause.

You cannot use this for code review or architectural analysis because it strictly performs evidence-based root-cause triage for broken software. Use separate code-review tools for auditing changes or architectural-analysis tools for structural risk assessment.

To troubleshoot a complex software integration failure, provide a symptom description to launch parallel investigators that collect evidence across concurrency and data flow. It compiles an evidence summary mapping root-cause findings to exact file paths and lines.

You should not use evidence-based investigation when researching open-ended options, assessing architectural health, or reviewing code quality. It is designed exclusively for diagnosing symptoms and tracing broken software to its root cause.

Root-cause investigation works for debugging API calls and integrations by collecting evidence from logs and code artifacts to trace failures. It dispatches specialized analysts to investigate symptoms from different angles like concurrency and data flow.