Snapshot v0.9.0

v0.9.0 binds 1,108 cases and 92 freshly executed report partitions, including 108 Swift cases. Its 4,252 outcomes comprise 1,309 reached, 1,472 not-reached, 579 inconclusive, 891 unsupported, and one runner-error. Incomplete and unsupported coverage remain visible and are never clean negatives.

Bifrost was measured at v0.12.0. All 33 prerequisite controls succeeded before 84 serial matrix groups ran. The single CodeQL Swift runner error retains its endpoint and resource-qualification diagnostics; it was not retried for a more favorable outcome. These counts span independent scorecards and are not a pooled analyzer ranking.

Timing qualification

Cold timing sidecars and fresh warm and invocation-overhead controls are bound to the same retained execution evidence. Missing timing remains missing. OS cache state was uncontrolled; desktop measurements establish no causal speedup.

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): e63dcc483d3000c046816c595811bf35ab494544fb6dd6fff2608d43e9bc676d

Freeze manifest: reports/freeze.json

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