askill
skill-validation-skill

skill-validation-skillSafety 100Repository

Validate skills using 84+ rules from AgentSkills OpenSpec, Nixtla, and Meta-Skill frameworks. Use when validating SKILL.md files or auditing skill quality. Trigger with /cpv-validate-skill.

1 stars
1.2k downloads
Updated 2/4/2026

Package Files

Loading files...
SKILL.md

Skill Validation Skill

Validates skill directories using a comprehensive set of 168+ validation rules extracted from:

  • AgentSkills OpenSpec (skills-ref library) - 44 rules
  • Nixtla Quality Standards (strict mode) - 52 rules
  • Meta-Skill Validation (8+1 Pillars, checklists) - 47 rules
  • Component Validators (multi-scale scoring, grading) - 25 rules

Quick Navigation

ResourcePurpose
references/validation-rules.mdComplete list of 168+ validation rules
references/frontmatter-schema.mdFrontmatter field requirements and validation
references/pillars-coverage.md8+1 Pillars validation for language skills
references/scoring-system.mdMulti-scale scoring (0-3) and letter grading (A-F)

Overview

This skill provides comprehensive validation for Claude Code skills according to multiple quality standards:

  1. Basic Validation - Structure, frontmatter, SKILL.md existence
  2. Quality Validation - Description quality, token budget, path formats
  3. Strict Mode - Nixtla required sections, voice checks
  4. OpenSpec Mode - AgentSkills strict field whitelist
  5. Pillars Mode - 8+1 Pillars coverage for lang-/convert- skills

Prerequisites

  • Python 3.12+ with pyyaml installed
  • The skill to validate must be a directory containing SKILL.md
  • For Pillars validation, skill name must start with lang- or convert-

Instructions

1. Run Basic Validation

uv run python scripts/validate_skill_comprehensive.py path/to/skill/

Output: Grade (A-F), issue counts, detailed results by category.

2. Run Verbose Validation (Show All Checks)

uv run python scripts/validate_skill_comprehensive.py path/to/skill/ --verbose

3. Run Strict Mode (Nixtla Quality Standards)

uv run python scripts/validate_skill_comprehensive.py path/to/skill/ --strict

Additional checks:

  • Required sections: Overview, Prerequisites, Instructions, Output, Error Handling, Examples, Resources
  • Description must include "Use when..." phrase
  • Description must NOT use first/second person
  • Instructions must have numbered step-by-step list

4. Run OpenSpec Mode (AgentSkills Strict Whitelist)

uv run python scripts/validate_skill_comprehensive.py path/to/skill/ --openspec

Additional checks:

  • Only 6 fields allowed: name, description, license, allowed-tools, metadata, compatibility
  • Name MUST match directory name
  • Extra fields cause MAJOR errors (not warnings)

5. Run Pillars Validation (Language Skills)

uv run python scripts/validate_skill_comprehensive.py path/to/skill/ --pillars

Validates 8+1 Pillars coverage:

  1. Module - import/export, namespacing
  2. Error - error handling, Result types
  3. Concurrency - async, threads, channels
  4. Metaprogramming - macros, decorators
  5. Zero/Default - null handling, Option types
  6. Serialization - JSON, encoding/decoding
  7. Build - package management
  8. Testing - test frameworks, assertions
  9. Dev Workflow/REPL (for REPL-centric languages)

6. Combine All Modes

uv run python scripts/validate_skill_comprehensive.py path/to/skill/ --strict --openspec --pillars --verbose

7. JSON Output (For CI/CD)

uv run python scripts/validate_skill_comprehensive.py path/to/skill/ --json

Output

Grade System (A-F)

GradeScore RangeMeaningAction
A90-100Production readyDeploy
B80-89Good, minor improvementsAddress minor issues
C70-79AcceptablePlan improvements
D60-69Reject, regenerateMajor rework required
F<60Fundamentally brokenRebuild

Exit Codes

CodeMeaning
0All checks passed (Grade A/B)
1CRITICAL issues found (Grade F)
2MAJOR issues found (Grade D)
3MINOR issues found (Grade C)

Severity Levels

LevelDefinitionAction
CRITICALSkill will not workMust fix before use
MAJORSignificant problemsCreate bug issue
MINORMay affect UX/qualityCreate enhancement issue
INFOSuggestionsOptional improvement
PASSEDCheck passedNo action needed

Validation Checklist

Copy this checklist and track your progress:

  • Run basic validation: uv run python scripts/validate_skill_comprehensive.py path/to/skill/
  • Fix all CRITICAL issues (Grade F blockers)
  • Fix all MAJOR issues (Grade D issues)
  • Run strict mode: add --strict flag
  • Address required sections (Overview, Prerequisites, etc.)
  • For lang-/convert- skills: run with --pillars flag
  • Verify 8+1 Pillars coverage (50%+ recommended)
  • Run with --verbose to review all checks
  • Achieve Grade B or higher before deployment
  • Document any intentionally skipped checks

Error Handling

Common Errors and Solutions

"SKILL.md not found"

  • Ensure the path points to a skill directory (not a file)
  • Check file is named SKILL.md (uppercase preferred)

"Malformed YAML frontmatter"

  • Ensure frontmatter starts and ends with ---
  • Validate YAML syntax (check for missing quotes, colons)

"Skill name must be lowercase"

  • Use kebab-case: my-skill-name not MySkillName

"Directory name must match skill name"

  • Rename directory or update name field to match

"Required section missing"

  • Add missing sections (only in --strict mode)
  • Sections: Overview, Prerequisites, Instructions, Output, Error Handling, Examples, Resources

"SKILL.md has X lines (max 500)"

  • Move detailed content to references/ subdirectory
  • Use progressive disclosure pattern

Examples

Example 1: Validate a Single Skill

$ uv run python scripts/validate_skill_comprehensive.py ./skills/my-skill/

======================================================================
Skill Validation: ./skills/my-skill/
======================================================================

Grade: B (85.2/100)

Summary:
  CRITICAL: 0
  MAJOR:    0
  MINOR:    3

Details:
  [Description Quality]
    [MINOR] Description should include 'Use when ...' phrase

  [Token Budget]
    [MINOR] SKILL.md has 520 lines (recommended: under 500)

----------------------------------------------------------------------
✓ Skill validation passed (Grade B)

Example 2: Validate with Strict Mode

$ uv run python scripts/validate_skill_comprehensive.py ./skills/lang-rust-dev/ --strict --pillars

======================================================================
Skill Validation: ./skills/lang-rust-dev/
======================================================================

Grade: A (94.5/100)

Pillars Coverage:
  ✓ Module: 1.0/1.0 - Full coverage with dedicated section
  ✓ Error: 1.0/1.0 - Full coverage (12 keyword occurrences)
  ~ Concurrency: 0.5/1.0 - Partial coverage (4 keyword occurrences)
  ✓ Metaprogramming: 1.0/1.0 - Full coverage with dedicated section
  ✓ Zero/Default: 1.0/1.0 - Full coverage (8 keyword occurrences)
  ✓ Serialization: 1.0/1.0 - Full coverage with dedicated section
  ✓ Build: 1.0/1.0 - Full coverage with dedicated section
  ✓ Testing: 1.0/1.0 - Full coverage (15 keyword occurrences)

----------------------------------------------------------------------
✓ Skill validation passed (Grade A)

Example 3: JSON Output for CI/CD

$ uv run python scripts/validate_skill_comprehensive.py ./skills/my-skill/ --json

{
  "skill_path": "./skills/my-skill/",
  "exit_code": 0,
  "overall_score": 92.5,
  "grade": "A",
  "counts": {
    "critical": 0,
    "major": 0,
    "minor": 1,
    "info": 2,
    "passed": 25
  },
  "pillar_scores": [],
  "category_scores": {},
  "results": [...]
}

Resources

Official Documentation

Reference Documents (in this skill)

Related Skills

  • plugin-validation-skill - Plugin-level validation
  • marketplace-validation-skill - Marketplace validation

See Also

  • /validate-skill command - Invoke this validation
  • skill-validation-agent - Agent for automated skill audits

Install

Download ZIP
Requires askill CLI v1.0+

AI Quality Score

96/100Analyzed 2/9/2026

An exceptionally high-quality meta-skill for validating other skills. It features comprehensive documentation, clear command-line instructions, detailed error handling, and well-structured examples. It follows industry standards like AgentSkills OpenSpec and Nixtla.

100
100
85
100
95

Metadata

Licenseunknown
Version-
Updated2/4/2026
PublisherEmasoft

Tags

ci-cdgithubgithub-actionsllmtesting