Quality Scoring Methodology

Every project in the PT-Edge directory receives a composite quality score from 0 to 100, computed from four equally-weighted dimensions. Scores are recalculated daily from live data.

Maintenance (0-25)

Is someone actively working on this?

  • Commit activity in the last 30 days (0-12 points)
  • Push recency — last push within 30d, 90d, 180d, or 365d (0-13 points)
  • Archived repositories score 0

Adoption (0-25)

Are people actually using this?

  • GitHub stars, logarithmic scale (0-10 points)
  • Monthly downloads from PyPI/npm, log scale (0-10 points)
  • Reverse dependents — other packages that depend on this (0-5 points)

Maturity (0-25)

Is this production-ready?

  • License present (8 points)
  • Published on PyPI or npm (9 points)
  • Repository age — older repos score higher (0-8 points)

Community (0-25)

Is there a community around this?

  • Fork count, logarithmic scale (0-15 points)
  • Fork-to-star ratio — higher ratio means more active community (0-10 points)

Quality tiers

Tier Score range What it means
Verified 70-100 Strong across all four dimensions. Actively maintained, widely adopted, production-ready.
Established 50-69 Solid in most dimensions. Reliable choice with some gaps (e.g., fewer downloads or fewer recent commits).
Emerging 30-49 Showing promise but not yet proven. May be new, lightly adopted, or missing packaging/licensing.
Experimental 10-29 Early stage. Limited adoption and community signals. May be a personal project or proof of concept.

Risk flags

Projects may carry risk flags that highlight specific concerns:

  • Archived — repository marked as archived by the owner
  • No License — no license file detected (legal risk for production use)
  • Stale 6m — no push in the last 180 days
  • No Package — not published on PyPI or npm
  • No Dependents — no other packages depend on this

Update frequency

Scores are recalculated daily. The pipeline pulls fresh data from GitHub, PyPI, npm, and other sources every morning (6am UTC), refreshes all materialized views, and regenerates the entire directory site. What you see on any page reflects the state of the ecosystem as of today.

Limitations

  • Stars and forks are popularity signals, not quality signals — a viral tweet can spike both without reflecting real utility
  • Download counts miss Docker-only or binary distribution channels
  • Single-contributor projects with low stars may be genuinely excellent but score low on adoption and community
  • The scoring formula weights all four dimensions equally — your priorities may differ