infer-scope-from-code

Infer capability scope by scoring knowledge-base capabilities against scan-index signals to generate a v1 scope.yaml. This eliminates manual, out-of-sync capability selection by mapping existing code to already-chosen domains and project profiles.

Evidence-aware triage for scope inference classifies capabilities into selected, deferred, and rejected categories based on scan-index evidence. It also identifies within-domain coverage gaps when knowledge-base inclusion rules fire but scan evidence is missing.

Generate scope.yaml by running scope inference during /codify using scan-index.json alongside prior domain-selection and project-profile proposals. The process applies bounded domain traversal via knowledge-base markdown listings to ensure deterministic artifact emission.

Yes, scope inference requires JSON-validated scan inputs and previously selected domains to function correctly. It restricts capability consideration to those domains and never invents evidence, ensuring deterministic and bounded selection.

Ensure traceability by generating a resolution trace and decision manifest alongside the scope.yaml artifact. These outputs explain grounding sources and confidence levels for each capability classification decision made during inference.

Scope inference produces coverage gaps when knowledge-base inclusion rules indicate a capability should exist within a domain, but the scan-index evidence is missing from the existing codebase. These gaps highlight areas requiring further development or documentation.