Trace Review

Use this skill when

• evaluating evidence quality
• debugging failures
• reviewing observability coverage
• deciding whether a run is diagnosable

Goals

• ensure execution can be reconstructed
• identify blind spots in tracing
• improve failure diagnosis and evaluation quality

Review criteria

• are the relevant identifiers present
• can the execution path be reconstructed
• are key transitions visible
• are side effects visible
• are retries distinguishable
• are errors surfaced clearly
• is the trace enough for future diagnosis

Output contract

Return:

• trace strengths
• missing trace elements
• impact of the blind spots
• recommended improvements