judge

Issue a Phase 1 verdict from experimental evidence and state files.

1|Updated Mar 23, 2026

Search Tags:#governance #evidence #drift #verdict #judge #phase-1

Ready to run

npx skills add https://github.com/LinkDry/Link-Research --skill judge-linkdry

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: judge
Source: https://github.com/LinkDry/Link-Research/tree/main/skills/judge
Command: npx skills add https://github.com/LinkDry/Link-Research --skill judge-linkdry

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

The Judge skill enforces a governance gate by producing a Phase 1 verdict only after drift is resolved and structured evidence exists, ensuring traceability and defensible decisions.

Core Features & Use Cases

Conservative, evidence-driven verdict: reads project state, artifact outputs (experiment-memory.md, STATE.md), and evidence bundles to produce a formal verdict.
Read/Write contract: writes judge-report.json, updates experiment state, and respects drift gating rules to guard against premature conclusions.
Governance posture: ensures visibility of decisions and supports subsequent Phase 2 handoffs when appropriate.

Quick Start

Provide the active experiment context and let Judge issue the Phase 1 verdict using the bound evidence.

Frequently Asked Questions about judge

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

FAQPage Schema

How do I issue a Phase 1 verdict after resolving experiment drift?▼

To issue a Phase 1 verdict after resolving experiment drift, you must provide canonical state from STATE.md and experiment-memory.md alongside structured outputs like analysis-report.json to generate the formal judge-report.json.

What is a drift-gate policy in experimental evidence evaluation?▼

A drift-gate policy in experimental evidence evaluation enforces conservative decisions by blocking formal verdicts until drift is resolved, requiring a bound anchor and structured outputs like config-snapshot.json before proceeding.

How do I automate Phase 1 governance reporting using experiment-memory.md?▼

Automate Phase 1 governance reporting by applying the Judge skill to experiment-memory.md and STATE.md, which enforces drift-gate rules and writes a formal judge-report.json to ensure defensible, traceable decisions.

Do I need a bound anchor and config-snapshot.json to run a Phase 1 evaluation?▼

Yes, you need a bound anchor and config-snapshot.json to run a Phase 1 evaluation, as the Judge skill applies only when drift is resolved and these structured evidence bundles are available for formal judgment.

When should I not use an automated governance gate for experiment verdicts?▼

You should not use an automated governance gate for experiment verdicts when drift remains unresolved or when structured outputs like analysis-report.json and canonical STATE.md files are missing from the project context.