Browse Store Rappterpedia GitHub

Public registry metrics

Every number below comes from a public, verifiable API — GitHub traffic and releases, the jsDelivr CDN, and the registry's own state files. No private analytics, no tracking pixel.

loading… · · ·

Traffic over time

daily clones, page views and CDN hits — accumulated past GitHub's 14-day window
Clones (downloads) Page views CDN hits

Critic score vs user score

Agent leaderboards

ranked across all registry agents

Publishers

Categories

Quality tiers

Rating spread

engine notes

Review angles

engine notes

Most downloaded files

jsDelivr CDN

Most visited pages

GitHub, 14-day window

Traffic sources

Releases

Where these numbers come from

Downloads = git clones + jsDelivr CDN file hits + GitHub release asset downloads. A clone pulls the whole registry; a CDN hit is a single agent file fetched by an installer or an agent. GitHub only exposes a rolling 14-day traffic window, so scripts/build_metrics.py snapshots it daily into state/metrics_history.json, deduplicated by date — the all-time total grows from the first day of tracking and never double-counts.

On downloads. The headline figure adds repository clones, CDN file hits and release asset downloads. Be aware that GitHub counts every actions/checkout as a clone, and this repository runs a dozen scheduled and push-triggered workflows — so clones, the largest component, are substantially self-generated. The estimated CI baseline is published as ci_clone_estimate alongside clones_excluding_ci_estimate in the snapshot. Agent file fetches is the narrowest and most trustworthy number on this page.

On critic scores. Only reviews written by a model count toward a critic score. When no model backend is reachable the panel falls back to a deterministic rubric — a line-count and regex heuristic — and those verdicts are reported separately and never published as a critic score. An agent no model has read shows as unrated rather than fresh.

On reviews. Only real user reviews from state/reviews.json — submitted through the GitHub issues API — are counted in the review total, the average rating, and the review leaderboards. The registry also carries engine-generated curator notes in state/curator_reviews.json; those are shown separately and labelled "engine notes", never mixed into the headline number.

On unique counts. GitHub reports uniques per 14-day window, not per day, so daily unique figures cannot be summed into a true all-time unique. The headline unique cloners figure is GitHub's own 14-day number; the "daily-unique sum" beside it is an upper bound, since one machine cloning on three days counts three times. Clone counts and CDN hits have no such caveat — they are plain event counts and do accumulate exactly.