Twin Score rubric
A Twin Score is two editorial numbers from 0 to 100, not a five-star widget with a made-up rater count. Scores are editorial 0-100 composites, not star averages and not user-count ratings. AggregateRating in JSON-LD is emitted only for these staff scores. JSON-LD AggregateRating, when present, is those staff scores scaled to 5, with reviewCount equal to editorial reviews — never a fake user total.
Human dimensions
| Dimension | Weight |
|---|---|
| Interface clarity | 14% |
| Onboarding | 12% |
| Documentation for humans | 12% |
| Collaboration | 12% |
| Support | 12% |
| Pricing fairness and clarity | 12% |
| Accessibility | 13% |
| Reliability | 13% |
Agent dimensions
| Dimension | Weight |
|---|---|
| API quality | 16% |
| MCP or tool surface | 10% |
| CLI | 12% |
| Structured docs | 14% |
| Auth for machines | 12% |
| Rate limits and quotas | 12% |
| Determinism | 12% |
| Skill packaging | 12% |
CSV and JSON live on the agents twin: twin-scores.json and the CSV sibling. License: CC BY 4.0.