Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →pollinations/abuse-detection
Detect and analyze abusive accounts on Pollinations. IP clustering, multi-signal scoring, ban recommendations.
Passed 20 of 22 test cases with the skill, and 5 of 22 without it.
and 20 more in the full report
15 fixed5 passed either way2 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 15 test cases the skill fixed. Same model, same prompt, with and without the skill.
In the composite abuse scoring query, what error rate percentage adds 15 points to the composite abuse score, and does an 80% error rate trigger 15 points or 10 points?
| Checked | Without skill | With skill |
|---|---|---|
| Assigns 15 points for error rates at or above 95%. | Missed | Passed |
| Assigns 10 points for error rates between 70% and 94%. | Missed | Passed |
The instructions your model reads when the skill loads.
sops and jq (the prod Tinybird read token comes from SOPS)Tinybird query pattern (prod workspace pollinations_enter, no row cap):
enter.pollinations.ai/observability/scripts/tb-prod.sh "SELECT ... FROM generation_event_v2 ... FORMAT JSONCompact"
Workspace: prod-only — staging has no real traffic.
tb-prod.sh --checkconfirms events are fresh. Don't switch or save credentials just to run an audit.
Quoting: Use double quotes for the SQL string. Use single quotes inside SQL. Avoid
!=with$'...'shell quoting (escaping issues) — preferNOT IN ('undefined', '')instead.
Six signals, each weighted independently:
| Signal | Max Points | Threshold | What it catches |
|---|---|---|---|
| IP cluster size | 30 | cluster * 0.15 | Multiple users sharing same IP hash |
| Zero pack spend | 15 | spend = 0 | No paid pack usage |
| Error rate | 15 | >= 95% (15pts), >= 70% (10pts) | Bots hammering failing endpoints |
| Moderation flags | 15 | >= 90% sexual (15pts), >= 50% (8pts) | NSFW generation bots |
| Disposable email | 15 | hotmail/outlook/proton + no spend | Random-string throwaway emails |
| IP rotation | 10 | >= 50 IPs (10pts), >= 20 (5pts) | Rotating through many exit IPs |
Score interpretation:
| Score | Action | False positive risk |
|---|---|---|
| 90-100 | Ban immediately | Very low |
| 70-89 | Ban after quick review | Low |
| 40-69 | Manual review needed | Medium |
| 10-39 | Monitor only | High — many legit users with NSFW or errors |
| 0-9 | Clean | N/A |
test cases passed, vs 5 / 22 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.