Data insights
How the ecosystem is growing
Three curves derived at build time from committed repository data: how plugins enter the registry, how much the verification pipeline actually tests each day, and how fast the ecosystem is moving to npm. Where the live dashboard shows the current state, this page looks at rates of change.
Data as of 2026-11-01 · registry 2026-11-02 · verification archive 2026-09-18
01 · Registry growth
Entries arrive in waves, then settle into a steady drip
28,578 entries have been indexed since 2026-08-13. Discovery comes in waves: the first full crawl peaked at 4,003 entries on08-18, a second crawl brought 2,656 on 08-21, and a re-discovery pass added 1,110 on 09-08. Since 09-09 the pipeline has settled into a steady state of roughly299–788 new entries per day.
Ecosystem apps (kind:app, 351 in total) almost all arrived with the first month's crawl — 239 of them landed in August, and since September they trickle in at about one per day. The manual submission queue accounts for129 entries so far.
By indexedAt (UTC). The last day is partial. The dashed line is the trailing 9-day average excluding the partial day. Discovery waves reflect pipeline runs, not organic bursts.
* partial weeks at the window edges. kind carries the current flag value, so this shows when apps were indexed — not the app share of the registry over time (the display-layer "ecosystem app" scope also includes unmounted plugins and has no day-level history).
02 · Verification throughput
Upstream releases drive the re-test waves
Every install-verification record — one plugin version against one dsh version — lands in the archive with a timestamp. Across 25 test days the pipeline has written132,877 of these test cells, peaking at 19,982 on09-06; 58,546 cells carry a real L5 web-run verdict.
The curve is driven by upstream: each dsh release (ticks) re-queues the whole registry, so the dip around 09-14 is the 0.1.6 switch-over rather than a slowdown — the wave that follows re-tests the registry against the new version.
Snapshot generated 2026-09-18 by scripts/collect-verification-throughput.ts fromsrc/data/verifications (manual refresh for now). Ticks mark upstream release dates; labels mark the first release of each minor line.
03 · npm adoption
A quarter of the registry now ships on npm
7,005 registry entries declare an npm package — up from339 in mid-August, an 11× increase in a month. As a share of cumulative registry entries, npm coverage climbed from ~11% through August to 24.7% on 11-01.
The climb is real, with two caveats: npm probe coverage itself expanded over this window (the 09-06 dip is probe fluctuation, not the ecosystem), and newly indexed entries arrive with npm attached at a much higher rate than the August baseline. A higher npm share matters beyond the number — install verification can test the published artifact instead of HEAD source, and npm is the faster lane in the verification ladder.
Numerator = entries in that day's npm snapshot. Denominator = cumulative entries indexed by that day (indexedAt, archived entries included). August ratios are dominated by expanding probe coverage — cross-period comparison is only meaningful from mid-September onward.