experiment-audit

Audit experiment integrity before claiming results. Uses cross-model review (external reviewer backend) to check for fake ground truth, score normalization fraud, phantom results, and insufficient scope. Use when user says "审计实验", "check experiment integrity", "audit results", "实验诚实度", or after experiments complete before writing claims.

Quality

88%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

77%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is highly actionable with a concrete workflow, validation checkpoints, and complete prompt/report templates, but it carries redundant reviewer-convention restatements and relies on external shared-references that are neither bundled nor cleanly signaled. Progressive disclosure is the weakest aspect given the absent bundle files.

Suggestions

Consolidate the reviewer-calling convention into one section; the codex/manual branch logic currently appears in Constants, Reviewer Calling Convention, and Step 2 with overlapping detail.

Extract the full audit prompt and report templates into dedicated reference files (e.g., references/audit-prompt.md, references/report-template.md) and link to them from the workflow steps to reduce inline bulk.

Verify or mark the shared-references/*.md links (external-cadence, reviewer-independence, reviewer-routing, review-tracing, integration-contract) as out-of-bundle dependencies so they are not read as navigable bundle files.

Dimension	Reasoning	Score
Conciseness	Mostly efficient and free of concept-over-explanation, but the reviewer-calling convention is restated across three sections (Constants, Reviewer Calling Convention, and Step 2) and the 'DO NOT summarize' directive is repeated, indicating it could be tightened.	3 / 5
Actionability	Provides a copy-paste-ready audit prompt with an exact A-F checklist, concrete config JSON, file-glob patterns, and complete markdown/JSON report templates, covering the common execution cases fully.	5 / 5
Workflow Clarity	The 4-step workflow is clearly sequenced with explicit PASS/WARN/FAIL validation checkpoints, error-recovery branches (manual-review unavailable -> stop and print install command), and integration feedback loops for a verdict-bearing batch operation.	5 / 5
Progressive Disclosure	Sections are well-organized, but no bundle files exist and the multiple shared-references/*.md links are buried inline mid-paragraph rather than cleanly signaled; large inline blocks (the full audit prompt, report templates) could live in separate files.	3 / 5
	Total	16 / 20 Passed

Description

100%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is exemplary: it concisely states the skill's purpose, enumerates concrete fraud-detection actions, and provides explicit bilingual trigger phrases covering both verbal and behavioral invocation conditions. It clearly distinguishes itself from other skills via its narrow experiment-integrity niche.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions covering the audit space comprehensively: 'check for fake ground truth, score normalization fraud, phantom results, and insufficient scope', matching the comprehensive-coverage anchor.	5 / 5
Completeness	Explicitly answers both what ('Audit experiment integrity... Uses cross-model review... to check for [four fraud patterns]') and when ('Use when user says... or after experiments complete before writing claims') with concrete trigger phrases.	5 / 5
Trigger Term Quality	Provides comprehensive natural trigger terms including bilingual synonyms ('审计实验', 'check experiment integrity', 'audit results', '实验诚实度') plus a behavioral trigger ('after experiments complete before writing claims').	5 / 5
Distinctiveness Conflict Risk	Occupies a clear niche (experiment-integrity auditing via cross-model review) with distinctive multilingual triggers unlikely to fire for unrelated skills, matching the minimal-conflict anchor.	5 / 5
	Total	20 / 20 Passed

Validation

81%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 13 / 16 Passed

Validation for skill structure

Criteria	Description	Result
allowed_tools_field	'allowed-tools' contains unusual tool name(s)	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
relative_links	Relative link issues: 1 suspicious	Warning

	Total	13 / 16 Passed

Repository: wanshuiyin/Auto-claude-code-research-in-sleep
Path: skills/experiment-audit/SKILL.md
Commit: 0472e53

Reviewed: 5 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.