product-adjudicator

Adjudicate multi-reviewer product evaluations into one prioritized, evidence-based release verdict.

1|Updated Aug 3, 2026
One-click install
npx skills add https://github.com/getyak/talent-signal --skill product-adjudicator-getyak
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: product-adjudicator
Source: https://github.com/getyak/talent-signal/tree/main/.agents/skills/product-adjudicator
Command: npx skills add https://github.com/getyak/talent-signal --skill product-adjudicator-getyak

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? Product reviews involving multiple specialist reviewers often produce conflicting scores, duplicated findings, and averaged-away safety concerns. This Skill turns independent specialist review packets into one traceable product decision with explicit vetoes, preserved disagreements, and owned next tests. ## Core Features & Use Cases - Panel orchestration: Freeze a review artifact, select the smallest sufficient reviewer panel from a lens map, and issue independent briefs so no reviewer sees another's verdict. - Contract validation: Validate specialist review and panel JSON against a machine-checkable contract using scripts/validate_review.py, and check panel skill structure with scripts/check_panel_skills.py. - Evidence-based adjudication: Apply jurisdiction, evidence-ranking, and veto rules so safety, privacy, and assessment-validity gates block releases instead of being averaged into scores. - Use Case: Before shipping an iOS screenshot-to-action feature, run recruiter-workflow, evidence-safety, and mobile-UX reviewers on the same frozen build, then produce a panel JSON with a release gate, top three findings, active vetoes, and retest conditions. ## Quick Start Use the product-adjudicator skill to run a review panel on build 42 of the iOS confirmation flow and produce a final release verdict.

Frequently Asked Questions about product-adjudicator

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I combine multiple AI reviewer opinions into one product decision?▼

Freeze the artifact, give each selected reviewer an independent brief with no access to other verdicts, validate their JSON packets, then adjudicate by evidence rank and domain jurisdiction. The output is a panel JSON with a release gate, top three findings, and veto resolutions rather than an averaged score.

How do I validate a product review JSON against a schema?▼

Run python3 scripts/validate_review.py path/to/review-or-panel.json. The validator auto-detects review versus panel documents, checks required fields, verdict and severity enums, score rules, and enforces that active vetoes require a blocked release gate.

When should a safety veto block a product release?▼

A veto blocks release when it covers wrong identity binding, unsupported facts presented as confirmed, unauthorized external writes, privacy failures, inaccessible consequential paths, deceptive candidate experience, or unvalidated candidate assessment. Active vetoes force release_gate to block and verdict to fail.

Can multiple AI personas count as independent reviewers?▼

No. Different persona prompts run by the same model are correlated and do not create statistical independence. The skill requires calibrating important evaluations against deterministic tests, blinded human review, and real user research.

What are the limitations of multi-reviewer panel adjudication?▼

Panels can create false rigor when the underlying artifact is thin, governance can become slower than the decision warrants, and strict lens maps can miss novel risks. The skill mandates abstention when evidence or required domain authority is missing.