| name | broadcast-enrich-metrics |
| description | Join raw platform metrics JSONL back to registry/manifest metadata and compute experiment scoring priors. |
Broadcast Enrich Metrics
Use this skill after broadcast-kit fetch-metrics has produced raw metrics JSONL and the user wants feedback records for learning loops, A/B tests, or topic selection.
Agent Instructions
- Input is raw metrics JSONL, usually from
state/<platform>/work/metrics/<account>/<date>.jsonl.
- Provide either a
publish-registry or one or more platform manifests so records can be attributed back to content metadata.
- Output is JSONL with
schema_version: broadcast.feedback.v0.
- Preserve the raw metrics inside the
metrics object.
- Use the computed
scores only as priors, not as final business truth.
- Use
--dry-run first to inspect match rate before writing output.
Command Template
broadcast-kit enrich-metrics \
--metrics <raw-metrics.jsonl> \
--registry <publish_registry.json> \
--manifest <optional-manifest.yaml> \
--output <feedback_enriched.jsonl> \
--dry-run
Remove --dry-run to write the enriched JSONL.
