session-deep-dive

You can validate scan heuristics by applying this Skill to high-signal session transcripts, which loads pre-computed metrics as hypotheses and confirms or refutes them using concrete transcript evidence.

Analyzing Claude session transcripts for friction points is best done by cross-referencing pre-computed metrics against real transcript data to generate per-session analysis reports highlighting evidence strength and actionable recommendations.

Yes, running a deep-dive session analysis requires the ccrider MCP to access transcript data, along with an existing metrics.jsonl file and reference templates for analysis and synthesis.

Yes, you can analyze the most recent high-signal session without an explicit ID by using the --last selector, which resolves the target session automatically from previous scan data.

Debug session-level issues using existing metrics by loading metrics.jsonl as hypotheses, validating them against actual transcript data, and generating cross-session synthesis reports to highlight friction points.

Automated heuristics for session transcript analysis can produce false positives, requiring qualitative deep-dive validation against actual transcripts to confirm evidence strength and generate accurate recommendations.