opus-clips-performance

You can generate per-clip engagement metrics by reading closed-loop post-manifests and joining each scheduled post to platform-native YouTube analytics. This produces detailed markdown reports showing clip scores, duration, and fetched engagement data.

YouTube analytics matching uses a tag-first approach via the [opus:<clip_id>] tag lookup, then falls back to a ±2 hour time-window match biased by closest duration. This robust joining ensures scheduled posts accurately connect to platform-native engagement data.

Currently, YouTube engagement is fully wired for analytics joining. For non-wired platforms, the system emits engagement:null with pending_task pointers in the manifest, leaving a stub for future analytics integration.

You need closed-loop post-manifests containing scheduled posts at least 24 hours old. Optional filters include project_id and since parameters to narrow the scope of manifests processed during bulk report generation.

The tool performs additive manifest updates keyed by scheduled_post scheduleId. It adds an engagement block when available or leaves a pending_task stub for platforms not yet wired, ensuring existing manifest data remains intact.

Yes, you can use the dry-run flag to preview per-clip engagement reporting and manifest backfill operations without writing changes. Platform restriction flags also allow you to limit processing to specific platforms.