scoring-checks

To add deterministic scoring checks, create a new check in src/scoring/checks/ that follows the default function signature, wire it into the orchestrator in src/scoring/index.ts, and validate the integration using TypeScript compilation and existing tests.

Deterministic scoring checks are standardized validation functions that ensure consistent and auditable scoring results by leveraging ScoringContext fingerprint data and git history. You need them when adding reliable, repeatable scoring logic to your pipeline.

You wire a new scoring check into the orchestrator by importing your Check[] returning function into src/scoring/index.ts. Reference shared values from src/scoring/constants.ts to maintain compatibility across the scoring pipeline.

Yes, you need existing tests to validate new scoring checks. Running TypeScript compilation and the existing test suite confirms that your newly wired check integrates correctly with the orchestrator and maintains system consistency.

The best way to keep scoring checks auditable is to leverage ScoringContext fingerprint data and git history within your check implementations. This approach ensures that every scoring result is traceable and deterministic.

You should use existing constants from src/scoring/constants.ts rather than custom values. Using the established constants ensures compatibility across the scoring pipeline and maintains consistency with other wired checks.