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

—

Does it follow best practices?

Impact

—

No eval scenarios have been run

Securityby

Passed

No known issues

Quality

Content

77%

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

A well-structured, highly actionable instruction skill with a clear workflow and explicit reviewer-backend calling conventions. Its main weakness is length — conceptual framing and integration pseudocode could be trimmed or externalized.

Suggestions

Tighten the 'Why This Exists' section: the four fraud patterns are already enumerated in the reviewer's audit checklist, so the body can reference them instead of restating.

Consider moving the 'Integration with Other Skills' pseudocode blocks into a shared-references file, since they describe behavior of other skills rather than this one.

Move the full reviewer audit prompt into a references/ file and inline only the variable substitutions, reducing the inline prompt block substantially.

Dimension	Reasoning	Score
Conciseness	Most content is operational (exact prompts, configs, templates), but at ~290 lines the conceptual 'Why This Exists' framing and the 'Integration with Other Skills' pseudocode add bulk that could be tightened.	2 / 3
Actionability	Gives concrete MCP tool names, exact config values, the full audit prompt, file glob patterns, and complete report templates — directly executable.	3 / 3
Workflow Clarity	A clearly sequenced four-step workflow (collect artifacts → send to reviewer → parse and report → summary) with explicit PASS/WARN/FAIL verdict checkpoints and a reviewer-independence boundary.	3 / 3
Progressive Disclosure	Well-signaled one-level-deep references to ../shared-references/* exist, but no bundle files are present in references/scripts/assets and the large inline audit prompt plus multiple templates are not split out.	2 / 3
	Total	10 / 12 Passed

Description

100%

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

A strong description: third-person imperative voice, concrete actions, explicit bilingual triggers, and a distinct niche. It answers both what and when without padding.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — 'check for fake ground truth, score normalization fraud, phantom results, and insufficient scope' — rather than vague language.	3 / 3
Completeness	Explicitly states both what ('Audit experiment integrity before claiming results' with named fraud patterns) and when ('Use when user says ... or after experiments complete before writing claims').	3 / 3
Trigger Term Quality	Provides natural, user-spoken triggers in two languages: '审计实验', 'check experiment integrity', 'audit results', '实验诚实度'.	3 / 3
Distinctiveness Conflict Risk	Occupies a clear niche (cross-model experiment integrity auditing) with distinctive triggers unlikely to fire for unrelated skills.	3 / 3
	Total	12 / 12 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
Commit: 82076e5

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.