session-replay

SKILL.md

This skill analyzes claude-trace JSONL files to provide insights into Claude Code session health, token usage patterns, error frequencies, and agent effectiveness. It complements the /transcripts command by focusing on API-level trace data rather than conversation transcripts.

| Empty file | Session had no API calls | Report "No data to analyze" | | Malformed JSON | Corrupted trace or interrupted write | Skip line, count in error report | | Missing fields | Older trace format | Use .get() with defaults | | Permission denied | File locked by another process | Clear error message, suggest retry |

| Encoding error | Non-UTF-8 characters | Report encoding issue |

Analice archivos JSONL de claude-trace para conocer el estado de la sesión, los patrones y la información procesable. Úselo para depurar problemas de sesión, comprender el uso de tokens o identificar patrones de falla. Fuente: rysweet/amplihack.

Ver original