Snapshot v0.9.2

This page presents the generated v0.9.2 results model. Counts and rates come from the frozen scorecards below; analyzer populations keep separate denominators. Inconclusive outcomes preserve incomplete proof; unsupported outcomes record declared coverage; runner-error outcomes record failed executions. None is converted into a negative finding.

Bifrost was measured at v0.15.1. CodeQL’s native Java, JavaScript, and Python profiles include runner errors caused by query-pack lookup failures. The case-variation note records that limitation and the other paired changes relative to v0.9.0.

No latency corpus is published for v0.9.2; raw durations remain in retained execution evidence, without a certified current timing comparison. The v0.9.0 latency page remains bound to its older pins and makes no timing claim about this snapshot.

Claim

Scope: release
Tracks: taint
Score tiers: calibration core language-extension modeling
Model profiles: benchmark-controlled tool-native
Exclusions: none

Bound evidence

Freeze ID (manifest SHA-256): 750e3b8c1275ee05b027017a1e993e84847e292344224976ca3447257ad15d69

Freeze manifest: reports/freeze.json

Continue to analyzers, languages, semantic templates, or case evidence.