aiskillstore/marketplace / skills/2389-research/judge
judge
VerifiedScoring framework for test-kitchen cookoff and omakase-off. Invoked at Phase 4 to evaluate implementations using 5-criteria scoring. Do not invoke directly - called by cookoff/omakase-off.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 5.7 KB |
| skill-report.json | config | 9.5 KB |
skill-report.json
View on GitHub{
"schema_version": "2.0",
"meta": {
"generated_at": "2026-07-06T01:49:17.924Z",
"slug": "2389-research-judge",
"source_url": "https://github.com/2389-research/claude-plugins/tree/main/test-kitchen/skills/judge",
"source_ref": "main",
"model": "codex",
"analysis_version": "3.0.0",
"source_type": "community",
"content_hash": "22ea97b2617e4a391197e61afec63e1084312cfcae1c7ea90d675d7049fa0258",
"tree_hash": "5d39f1b9ca14b8cdb445b6aab1f84891e3beef47d5a0f2ca9c6ae543094e26ff",
"provenance": {
"model_requested": "gpt-5.5:high",
"model_effective": "codex:gpt-5.5:high",
"fallback_chain": [
{
"agent": "codex",
"model": "gpt-5.5",
"effort": "high",
"outcome": "success"
}
],
"analysis_version": "3.0.0"
}
},
"skill": {
"name": "judge",
"description": "Evaluate competing code implementations with a structured five-criterion scoring worksheet for test-kitchen cookoff and omakase workflows.",
"summary": "Scores competing implementations and selects a winner with gates, rationale, and trade-off notes.",
"icon": "📦",
"version": "1.0.0",
"author": "2389-research",
"license": "MIT",
"category": "coding",
"tags": [
"code-review",
"implementation-scoring",
"test-kitchen",
"quality-rubric",
"comparison"
],
"supported_tools": [
"claude",
"codex",
"claude-code"
],
"risk_factors": [
"external_commands"
]
},
"security_audit": {
"risk_level": "safe",
"is_blocked": false,
"safe_to_publish": true,
"analysis_status": "ok",
"summary": "The two static findings are false positives caused by Markdown template text. The fenced block is an output format example, and the trade-off placeholder does not request system reconnaissance. No prompt injection, data exfiltration, or malicious intent was found in SKILL.md.",
"remediation": [],
"risk_factor_evidence": [
{
"factor": "external_commands",
"evidence": [
{
"file": "SKILL.md",
"line_start": 18,
"line_end": 129
}
]
}
],
"static_findings": [
{
"id": "external_commands:SKILL.md:18:ruby-shell-backtick-execution",
"category": "external_commands",
"severity": "medium",
"pattern": "Ruby/shell backtick execution",
"file": "SKILL.md",
"line_start": 18,
"line_end": 129,
"snippet": "```markdown"
},
{
"id": "blocker:SKILL.md:128:system-reconnaissance",
"category": "blocker",
"severity": "low",
"pattern": "System reconnaissance",
"file": "SKILL.md",
"line_start": 128,
"line_end": 128,
"snippet": "[What the other implementations did better]"
}
],
"finding_verdicts": [
{
"id": "external_commands:SKILL.md:18:ruby-shell-backtick-execution",
"verdict": "false_positive",
"confidence": 0.95,
"reason": "Line 18 opens a fenced Markdown example for the required output format. It is not Ruby backtick execution and does not instruct the agent to run an external command."
},
{
"id": "blocker:SKILL.md:128:system-reconnaissance",
"verdict": "false_positive",
"confidence": 0.93,
"reason": "Line 128 is a placeholder asking the judge to summarize trade-offs between candidate implementations. It does not ask for host, network, account, or environment reconnaissance."
}
],
"semantic_findings": [],
"critical_findings": [],
"high_findings": [],
"medium_findings": [],
"low_findings": [],
"dangerous_patterns": [],
"files_scanned": 1,
"total_lines": 178,
"audit_model": "codex",
"audited_at": "2026-07-06T01:49:17.924Z",
"agent_auto_install_policy": "allowed",
"manual_install_policy": "allowed"
},
"content": {
"user_title": "Score Code Implementation Cookoffs",
"value_statement": "Comparing multiple AI-built implementations can be inconsistent and slow. This skill provides a fixed rubric for Claude, Codex, and Claude Code reviews.",
"seo_keywords": [
"Claude",
"Codex",
"Claude Code",
"AI code review",
"implementation scoring",
"test kitchen judge",
"code comparison",
"software evaluation",
"quality rubric",
"cookoff scoring"
],
"actual_capabilities": [
"Produces a required Markdown worksheet for comparing candidate implementations.",
"Scores fitness, complexity, readability, robustness, and maintainability.",
"Applies hard gates for large fitness gaps and critical flaws.",
"Checks feasibility risks such as unbounded memory, missing pagination, and no recovery.",
"Selects a winner with rationale and trade-off notes."
],
"limitations": [
"Does not run tests or inspect files without context from the host agent.",
"Requires reviewer judgment, so scores can vary between evaluators.",
"Focuses on implementation quality, not security certification or product strategy.",
"Uses a verbose fixed format that may be too detailed for quick reviews."
],
"use_cases": [
{
"title": "Compare AI Code Variants",
"description": "Score several generated implementations of the same design and choose the strongest result.",
"target_user": "AI coding workflow user"
},
{
"title": "Review Engineering Cookoffs",
"description": "Use one rubric to compare competing solutions during an internal implementation challenge.",
"target_user": "Engineering lead"
},
{
"title": "Document Selection Decisions",
"description": "Create a repeatable decision record with scores, gates, winner rationale, and trade-offs.",
"target_user": "Technical reviewer"
}
],
"prompt_templates": [
{
"title": "Score Two Implementations",
"scenario": "Beginner comparison after both implementations are in context.",
"prompt": "Use the judge skill to compare impl-1 and impl-2. Fill every worksheet section and select the winner."
},
{
"title": "Judge A Cookoff",
"scenario": "Standard review for three implementations of one design.",
"prompt": "Use the judge skill for impl-1, impl-2, and impl-3. Apply all five scoring criteria and hard gates."
},
{
"title": "Evaluate Omakase Variants",
"scenario": "Advanced comparison of different solution approaches.",
"prompt": "Use the judge skill to compare variant-a and variant-b. Treat fitness differences as real product fit differences."
},
{
"title": "Audit A Close Decision",
"scenario": "Advanced review when scores are close or trade-offs are subtle.",
"prompt": "Use the judge skill and explain any tied or close criteria. Make the winner rationale evidence-based and specific."
}
],
"output_examples": [
{
"input": "Compare two implementations that both pass tests but differ in complexity.",
"output": "A worksheet with passing gates, five criterion scores, total scores, and a winner chosen for lower justified complexity."
},
{
"input": "Judge three variants where one handles scale poorly.",
"output": "A scorecard that flags feasibility risks, lowers robustness for the weak variant, and explains the final winner."
},
{
"input": "Choose between two approaches with different product fit.",
"output": "A decision record that applies the fitness gate and documents what the losing approach did better."
}
],
"best_practices": [
"Provide all candidate implementations and the original requirements before invoking the skill.",
"Use evidence from tests, code structure, and realistic user scenarios for each score.",
"Keep scores as integers from 1 to 5 and fill every required section."
],
"anti_patterns": [
"Do not use the skill before the implementations and requirements are available.",
"Do not skip the worksheet and provide only a winner.",
"Do not award half points or change the scoring scale."
],
"faq": [
{
"question": "When should I use this skill?",
"answer": "Use it when two or more implementations need a structured comparison and a clear winner."
},
{
"question": "Does the skill execute tests?",
"answer": "No. It records test results supplied in context and uses them during scoring."
},
{
"question": "Can it compare different designs?",
"answer": "Yes. The omakase guidance supports different approaches and treats fitness gaps as product fit differences."
},
{
"question": "What criteria does it score?",
"answer": "It scores fitness for purpose, justified complexity, readability, robustness and scale, and maintainability."
},
{
"question": "Can I change the output format?",
"answer": "The skill is designed for a fixed worksheet. Changing the format reduces consistency across reviews."
},
{
"question": "Is this a security audit tool?",
"answer": "No. It can notice robustness issues, but it is not a dedicated security assessment."
}
]
},
"file_structure": [
{
"name": "SKILL.md",
"type": "file",
"path": "SKILL.md",
"lines": 178
}
]
}
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| finishing-a-development-branchobra/superpowersUse when implementation is complete, all tests pass, and you need to decide how to integrate the work | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixes | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation code | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| skill-creatoranthropics/skillsCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |