review-judgment

You can automate a single-shot judgment by providing the prosecution ledger and defense responses, which triggers evidence verification and a scoring model to emit a final, machine-readable ruling.

A machine-readable judgment payload is a structured output block that contains the final ruling and score summary, designed for automated pipelines to consume after evaluating prosecution and defense evidence.

To map evidence severity to points, you need a scoring model that assigns numerical values to findings, allowing the judgment process to quantify prosecution and defense claims before generating a final ruling.

The single-shot judgment workflow uses a low-deferral design, meaning it provides a final ruling with no further rounds when evidence is conclusive, rather than supporting indefinite deferral.

Yes, you can run evidence-backed judgments without external dependencies because the Skill operates independently to read cited evidence directly and output structured judge-rulings payloads.

The best way to verify cited evidence is through independent verification, reading the referenced evidence directly from the prosecution and defense inputs before applying the scoring model to issue a decisive ruling.