What problem does it solve?
Error tracking issues in PostHog can have hundreds of linked session recordings, most of which are crash-only fragments or duplicates. Manually picking a useful replay from that list is slow and often yields sessions with no pre-error context.
Core Features & Use Cases
- Candidate discovery: Queries exception events by issue fingerprint to collect up to 20 recent session IDs where the error occurred.
- Quality-based ranking: Filters out sessions under 10 seconds or over 1 hour, then ranks by duration sweet spot (2-15 minutes), active time ratio, activity score, and recency.
- Contextual presentation: Returns the best recording with a watch link, pre-error page journey, and error frequency for that session.
- Optional AI summary: Uses Replay Vision summarizer scanners (check-then-scan) to generate a narrative summary of the selected session.
- Use Case: A developer sees a spike in an error tracking issue and asks for a replay; the skill returns the single most informative recording showing what the user did before the exception.
Quick Start
Find the most informative session replay for error tracking issue <issue_id> and summarize what the user was doing before the error.
