Trust Methodology v1

How A Screener Scores Skills

A Screener is a marketplace first. Rankings are based on verified telemetry and verification runs. Sponsored placement never changes organic trust scores.

Back to Screener

Formula

trust = 0.35 * reliability + 0.30 * security + 0.20 * performance + 0.15 * adoption

  • Reliability: successful executions over total executions.
  • Security: penalty model based on detected risk events.
  • Performance: latency efficiency benchmarked over time.
  • Adoption: quality-adjusted usage with diminishing returns.

Decay and Freshness

  • No successful run in 14 days: score reduced by 10%.
  • No successful run in 30 days: score reduced by 25%.

Data Quality Rules

  • Signed telemetry required for ingestion.
  • Events older than 24 hours are rejected.
  • Duplicate session events are ignored.

Disclaimer

Scores are decision support, not guarantees. Always review permissions, code, and runtime behavior before production use.