Content quality
Route: /content
Scenario: AI-generated copy, release notes, or knowledge snippets scored poorly or got user flags — you need a queue of fixable assets.
The Content Quality Debug Station lists assets that need review:
| Reason | Meaning |
|---|---|
low_judge_score | The score the source app sent (judge_score) is below its threshold. A source may send a vote score here rather than an AI judge grade, so the console calls it “Low score” |
user_flag | End users flagged the content |
low_star_rating | Aggregate star rating is low |
high_downvote_ratio | Downvotes dominate |
Open a row (/content/:id) to inspect scores and trigger regeneration. Stats
strip and readout follow the standard PagePosture chrome.
To clear noise in bulk, tick rows or filter the list (status, reason, content type, source, score band), then choose Dismiss selected or Dismiss all matching. The dialog states the exact number of rows and needs a reason, which goes to the audit log. One run dismisses at most 10,000 rows. A dismissed row stays dismissed when the source re-sends the same item; only new flags, new downvotes or a clearly lower score reopen it.
Related pages
- Judge dashboard — quality grades that feed this queue
- Releases — changelog / release-note surfaces