Scorecard
Comparative scoring is represented honestly: priority scoring is implemented, while the spec rubric is a supervised manual stage unless a run artifact supplies evidence.
Scoring Gap Disclosure
`lib/scoring.js` scores finding priority using KEV, EPSS, CVSS, asset, exposure, and reachability signals. The model-comparison rubric in the spec requires an Opus 4.8 scorer pass and should not be faked from seeded values.
Fable 5
manual required
Primary Hunt + Validate
claude-fable-5
Claude Opus 4.8
supervised manual
Scorer of record + gapfill
claude-opus-4-8
GPT-5.5
manual required
Cross-validation
gpt-5.5