Install · Vol.3
14,908 plugins tested: 14,123 install (L4 94.7%), 7,799 pass the L5 runtime smoke (71.26%, 3,964 eco-apps excluded).
Vol.3 · frozen 2026-09-15 · dsh 0.1.5rc214,908Tested pluginsregistry 15,132 · legacy denominator includes eco-apps
7,799Runtime-verified (L5)dsh 0.1.5rc2
71.26%L5 pass ratepass / (tested − apps 3,964)
3,145L5 failures88% one cause · see failure profile
3,964L5 eco-appsexcluded from ratios (legacy)
14,123L4 installs−109 install failures
Install funnel — registry to runtimeVol.3
Registry entries
15,132
100% · everything indexed
Tested
14,908
100.00% · legacy caliber · 3,964 eco-apps excluded
L4 installs
14,123
94.7% · −109 install failures
L5 runsruntime-verified
7,799
71.26% · −6,324 installed but never ran
Failure profile3,145failures
Plugin tree load failure2,752 · 88%
Boot failed113 · 4%
Missing from loader inventory70 · 2%
Duplicate loader entry id49 · 2%
HTTP not served39 · 1%
L5 pass rate · by version70.0%→ 72.0%
0.1.2a10.1.3a10.1.5rc2
passedpass rate
L4 installs vs L5 runs94.7%→ 71.3%
0.1.2a10.1.2rc10.1.3a10.1.3a20.1.5rc2
L4 installsL5 runs
Two bars per dsh version: left = installed in the sandbox (L4), right = actually runs (L5). The gap is "installs but will not run".
Channels × outcomes14,908tested
One row per install channel (npm package / GitHub source); the coloured segments are the passed and failed shares. npm releases are not re-tested from source.
Caliber: pass rate = L5 pass ÷ (tested − captured apps / unknown); per-version denominators = pass + fail; L1-L3 static spec ≈100% (folded into the funnel footnote) · per-plugin install record /data/install/<id>.json · historical volumes render from frozen snapshots (Vol.1-3 predate the three-ledger caliber)