skill-check（技能检查）

SkillCheck (Free)

Check skills against Anthropic guidelines and the agentskills specification. This file contains Free tier validation rules.

Want deeper analysis? Upgrade to Pro for anti-slop detection, security scanning, token optimization, WCAG compliance, enterprise checks, and Eval Kit (auto-generated test prompts for your skills).

Prerequisites

• Any AI assistant with file Read capability (Claude Code, Cursor, Windsurf, Codex CLI)
• Works on any platform (Unix/macOS/Windows)
• No special tools required (Read-only)

How to Check a Skill

1. Locate: Find target SKILL.md file(s)
2. Read: Load the content
3. Validate: Apply each rule section below
4. Report: List issues found with severity and fixes

Free Tier Validation Rules

Apply these checks in order. Each subsection defines patterns to match and issues to flag.

1. Frontmatter Structure

Every SKILL.md must start with YAML frontmatter between --- markers.

Required Fields

Frontmatter Security

Check 1.9-xml-in-frontmatter (Critical): Frontmatter values must not contain XML angle brackets (< or >). Frontmatter appears in Claude's system prompt; angle brackets could enable prompt injection.

Detection: Scan all frontmatter string values (name, description, compatibility, etc.) for < or > characters.

Fix: Remove angle brackets from frontmatter. Use plain text descriptions. Markdown formatting and XML tags are fine in the SKILL.md body.

Optional Fields (Spec)

Fields defined in the agentskills.io specification:

Claude Code Extensions

Recognized by Claude Code but not part of the agentskills.io spec. Other agents may ignore these fields.

Community Extensions

Not part of any spec. Used by community tools and registries.

Category Validation

Note: category is a Claude Code extension, not part of the agentskills.io spec. Do not flag a missing category field. Only validate format if present.

Format: String or array of strings, lowercase letters, numbers, and hyphens only.

Pattern: ^[a-z][a-z0-9-]*[a-z0-9]$ (same rules as skill name)

Common categories: development, productivity, data, automation, writing, design, security, devops, api, testing, documentation, legal, financial, marketing, ai-ml

Artifact Passing Validation

Note: produces and consumes are Claude Code extensions for inter-skill artifact passing. Do not flag missing fields. Only validate format if present.

Format: Comma-separated list of artifact type names. Each type must be lowercase with hyphens only.

Known artifact types: content-brief, knowledge-note, reading-log-entry, sift-article

Check 1.10-artifact-types (Warning): If produces or consumes contains a type not in the known list above, flag as a warning (not an error). New types are valid but should be registered in rules/artifact-passing.md.

Check 1.11-consumes-without-tools (Warning): If a skill declares consumes: but its allowed-tools does not include Read or Glob, flag as a warning. Consuming artifacts requires reading files.

Name Validation

Pattern: ^[a-z][a-z0-9-]*[a-z0-9]$

Naming suggestions: Avoid generic terms that don't describe what the skill does: helper, utils, tools, misc, stuff, things, manager, handler. Product-specific terms (claude, anthropic, mcp) are allowed but may limit portability across agents.

Description Validation

Must contain:

1. WHAT: Action verb explaining what skill does
2. WHEN: Trigger phrase for when to use it
3. Key capabilities (recommended): Specific tasks or file types handled

Recommended structure: [What it does] + [When to use it] + [Key capabilities]

Action verbs: Create, Generate, Build, Convert, Extract, Analyze, Transform, Process, Validate, Format, Export, Import, Parse, Search, Find

WHEN triggers: "Use when", "Use for", "Use this when", "Invoke when", "Activate when", "Triggers on", "Auto-activates", "Run when", "Applies to", "Helps with"

allowed-tools Validation

Both space-separated and YAML list formats are valid:

Directory Structure Validation

Skills can include optional subdirectories per the agentskills spec:

Check 1.10-readme-in-folder (Warning): Skill folder must not contain a README.md file. All documentation goes in SKILL.md or references/. For GitHub distribution, place the README at the repo root, outside the skill folder.

Detection: Use Glob to check if {skill-dir}/README.md exists.

Skill path formats supported:

• Standard: ~/.claude/skills/{skill-name}/SKILL.md
• Namespaced: ~/.claude/skills/{namespace}/{skill-name}/SKILL.md

Namespace support: Namespaces allow organizing skills by source (personal, team, project):

• ~/.claude/skills/internal/weekly-reports/SKILL.md
• ~/.claude/skills/shared/code-review/SKILL.md

Directory name must match skill name: The parent directory name must exactly match the name field in frontmatter.

2. Naming Quality

Names should be descriptive compounds, not single words.

Length Guidelines: Minimum 3 chars, optimal 10-30 chars, maximum 64 chars.

Anti-Pattern Format Lint

Check 2.8-antipattern-format (Suggestion): When a skill documents anti-patterns (sections with headers matching "anti-pattern", "what not to do", "avoid", "common mistakes", "bad practices", "pitfalls"), the content should use structured formats (tables or bullet lists) rather than wall-of-text prose.

Fires when:

• Anti-pattern section has a long prose line (100+ chars) containing don't/avoid/never
• Anti-pattern section has 3+ prose lines with 3+ avoidance directives

Does NOT fire when:

• Section already uses tables (| col | col |) or bullet lists (- item)
• Section header doesn't match anti-pattern keywords

You should avoid using global variables because they create hidden dependencies and you should never skip error handling because it leads to silent failures in production and makes debugging very difficult.

reason: Wall-of-text prose; restructure as table or bullet list

3. Semantic Checks

Validate logical consistency and clarity of skill instructions.

Contradiction Detection

Flag conflicting instructions that simultaneously require and forbid the same action.

Ambiguous Terms

Flag vague language that should be more specific. Terms like "multiple items" or "correct settings" lack precision. Use exact counts or specific criteria instead.

Exceptions (not flagged):

• Terms inside code blocks or blockquotes
• Content in example/usage/pattern sections
• Before/After and correct/incorrect comparison lines
• Terms followed by qualifiers (e.g., "some specific files")

Output Format Specification

Skills that mention output should specify format with concrete examples.

Detection:

• Skill mentions "output/returns/produces" without ## Output section
• Has output section but lacks code blocks, JSON, or tables

Returns JSON:

{
  "status": "success",
  "items": ["a", "b", "c"]
}

Returns the processed data.

reason: No concrete format example

Wisdom/Platitude Detection

Check 4.6-wisdom-platitude (Suggestion): Detects generic advice ("wisdom") that lacks actionable content. Skills should contain concrete instructions, not motivational prose.

Three detection layers:

1. Opener patterns: Lines starting with wisdom phrases like "Remember that", "It's important to", "Keep in mind that", "Think about", "Never forget that", "Always keep in mind", "Consider the importance of"
2. Platitude structures: Mid-line "[noun] is essential/crucial/important to [noun]" patterns
3. Vague imperatives: "Ensure quality", "maintain standards", "strive for best practices"

Exceptions (not flagged):

• Content inside code blocks or blockquotes
• Content in example/usage/pattern sections
• Before/After and positive/negative comparison lines

reason: Generic advice; replace with specific, actionable directives

Description Trigger Style

Check 4.8-description-trigger-style (Suggestion): The description field should read as a trigger condition, not a capability summary. Claude scans descriptions to decide "is there a skill for this request?" A trigger-oriented description activates correctly; a summary-oriented one gets overlooked.

Detection: Description opens with summary patterns instead of trigger patterns:

• Summary openers (flag): "This skill", "A tool that", "Provides", "Offers", "Handles", "Manages", "Enables"
• Trigger openers (pass): "Use when", "Generate", "Build", "Convert", "Extract", "Validate", "Run", "Create", "Find", "Search"

Does NOT fire when:

• Description has a summary opener but also contains a WHEN trigger phrase later ("Handles X. Use when user says Y" is fine)
• Description starts with an action verb

Severity: Suggestion

Fix: Rewrite to lead with action verb and include "Use when" clause. The description is a routing instruction, not documentation.

Railroading Detection

Check 4.9-railroading (Suggestion): Skills should give Claude information and context, not dictate exact sequences. Excessive prescriptive language reduces Claude's ability to adapt to the user's actual situation.

Detection: Count prescriptive phrases outside example blocks, code blocks, and anti-pattern sections:

• "you must always", "always do exactly", "never deviate", "follow these exact steps", "do not change this", "this is the only way", "you are required to"

Fires when: 5+ prescriptive phrases in non-example content.

Does NOT fire when:

• Prescriptive language is inside code blocks, blockquotes, or example tags
• Prescriptive language is in anti-pattern/gotchas sections (where it describes what NOT to do)
• Skill is a safety-critical skill (security, compliance) where strict instructions are warranted

Severity: Suggestion

The config file lives at ~/.config/myskill/config.json. If missing, Claude asks the user for the required values.

Common options:

• output_dir: where reports land (default: current directory)
• format: json or markdown (default: markdown)

You must always check the config file at ~/.config/myskill/config.json. You must never deviate from this path. Always do exactly as follows: first read the config, then validate every field. You must always use JSON format. Never change the output directory.

reason: 5 prescriptive phrases; give Claude the information and let it adapt

Misplaced Routing Content

Check 4.4: Body contains trigger conditions that belong in the description field.

Detection: Body contains a heading matching ## When to Use or ## When to Use This Skill, or body text contains routing phrases like "Activate when user", "Trigger this skill when", "Use this skill when".

Problem: The skill body loads only AFTER the Skill tool is invoked. Trigger conditions placed here don't influence routing decisions. Claude reads the description field during routing; that's where "Use when" patterns, trigger keywords, and example phrases belong.

Severity: Warning

Fix: Merge unique trigger content from the body section into the description field, then remove the redundant body section.

When to Use

Activate when user says "weekly report" or "generate summary".

reason: Trigger phrases are invisible during routing; they only load after invocation

How to Use

1. Provide data source path
2. Run report generation

4. Quality Patterns (Strengths)

Recognize positive patterns in skills. These are reported as "strengths" rather than issues.

8.1 Has Example Section

Skills with ## Example, ## Usage, or <example> tags demonstrate expected behavior clearly.

Strength: "Skill includes example section"

8.2 Has Error Handling

Skills documenting limitations, error cases, or edge cases set correct expectations.

Patterns detected: ## Error, ## Limitation, does not support, will fail if

Strength: "Skill documents error handling or limitations"

8.3 Has Trigger Phrases

Description includes activation triggers (Use when, Triggers on, Applies to, etc.)

Strength: "Description includes activation triggers"

8.4 Has Output Format

Skills specifying output format with concrete examples (code blocks, JSON, tables).

Strength: "Skill specifies output format with examples"

8.5 Has Structured Instructions

Skills using numbered steps or clear workflow sections.

Strength: "Skill uses structured instructions"

8.6 Has Prerequisites

Skills documenting setup requirements or dependencies.

Strength: "Skill documents prerequisites"

8.7 Has Negative Triggers

Description includes scope boundaries that prevent over-triggering.

Patterns detected: "Do NOT use for", "Not for", "Don't use when", "not intended for", "Do not use for"

Strength: "Description includes negative triggers to prevent over-triggering"

8.9 Has Gotchas Section

Skill documents common failure points, edge cases, or footguns. Per Anthropic's internal skill design guidance, gotchas sections are "the highest-signal content in any skill."

Patterns detected: Section headers matching "Gotchas", "Pitfalls", "Common Mistakes", "Watch Out", "Known Issues", "Caveats", "Footguns", "Common Errors"

Strength: "Skill documents gotchas or common failure points"

Pro Tier Features

Available with SkillCheck Pro:

Error Handling

If validation stalls on large files (1000+ lines), break the skill into smaller modules or check frontmatter syntax first.

Severity Levels

Check IDs Reference

5. Design Pattern Classification

SkillCheck classifies each skill into one of five design patterns from the Google ADK taxonomy:

• Reviewer: evaluates output against criteria (e.g., skill-check, code-review)
• Generator: produces structured artifacts from templates (e.g., linkedin-post, brand-assets)
• Inversion: asks user questions before acting (e.g., grill-me, feature-scoping)
• Pipeline: chains multiple steps with checkpoints (e.g., sift, tapestry)
• Tool Wrapper: wraps an API with context-aware instructions (e.g., lmwtfy)

Check 19.1-pattern-detected (Strength): Pattern identified. Reports primary pattern and any secondary patterns. Most skills are hybrids.

Pro checks 19.2-19.7: Validate pattern-specific requirements. A Reviewer without criteria, a Generator without output format, an Inversion without question framework, a Pipeline without stages, or a Tool Wrapper without API boundaries.

Autonomy Design

Stop condition: Stop after reporting all issues for the target SKILL.md. Do not iterate or re-check unless the user requests it.

Budget: Maximum attempts: 1. Single SKILL.md validation completes in one pass with maximum retries: 0. No looping.

Idempotency: Safe to re-run. Same input produces the same severity counts and issue list.

Composability

Input/Output contract:

• Input: expects: path to a SKILL.md file or directory containing one.
• Output: returns: structured report with overall score (0-100), issue list (check_id, severity, line, message, fix), and passed count.

Output structure:

# SkillCheck Report: {skill-name}
## Summary (score, issue/strength counts)
## Issues (check_id, severity, line, message, fix per issue)
## Strengths (check_id, detail per strength)

Mode: Runs identically standalone or in composition with other skills. Self-contained, read-only, no side effects.

Platform Requirements

• Code Execution: No
• File Creation: No
• Network Access: No
• Required tools: Read, Glob

Observability

Confidence level: High confidence on structural and pattern-based checks. Semantic checks (contradiction detection, wisdom/platitude) are heuristic; ~5% false positive rate.

Progress: Step 1 of 4: Parse. Step 2 of 4: Validate. Step 3 of 4: Score. Step 4 of 4: Report.

Phase Boundaries

Phase 1: Parse — Read file, extract frontmatter. Phase 2: Validate — Run all Free tier checks. Phase 3: Score — Calculate overall score. Phase 4: Report — Format and return results.

Gotchas

• YAML frontmatter with multi-line descriptions may trigger false positives on line-based checks. Use a single-line description or pipe (|) syntax.
• Skills using non-standard frontmatter fields (custom extensions) get informational notices, not errors. These are expected and safe to ignore.
• Check 4.6 (wisdom/platitude detection) has a ~5% false positive rate on technical content that uses advisory language. If flagged incorrectly, the content is fine; the heuristic is conservative.
• The "misplaced routing" check (4.4) fires on body content that contains trigger phrases. If your body has a "Quick Start" section showing example invocations, wrap them in code blocks or <example> tags to avoid false positives.

Upgrade to Pro

Get the complete suite at getskillcheck.com: Go binary for CI/CD, MCP server for IDE integration, all Pro checks, auto-fix, and badge generation.

Skills can also be distributed via the /v1/skills API endpoint. See agentskills.io for the specification.