Skill-Audit for Openclaw

A technical linter and quality assurance tool designed to scan, score, and optimize your AI agent skill library.

bill492
v1.0.0
Mar 18, 2026
0
924
0

Install & Download

1. ClawHub CLI

The fastest way to install a skill directly from the registry.

npx clawhub@latest install unblocklabs-skill-audit

2. Manual Installation

Copy the skill folder to one of these locations

Global
~/.openclaw/skills/
Workspace
<project>/skills/

Priority: Workspace > Local > Bundled

3. Prompt Installation

Copy this prompt to OpenClaw to install it automatically.

Help me install unblocklabs-skill-audit using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).

Prefer to download?

Get the raw skill files in a ZIP archive.

What is Skill-Audit?

Skill-Audit is a specialized utility designed to maintain the hygiene and performance of your AI agent ecosystem. By scanning global, workspace, and project directories, it ensures that every skill adheres to high-signal documentation standards, such as trigger-oriented descriptions and the inclusion of critical Gotchas sections. This tool is essential for developers who want to scale their Openclaw Skills while avoiding technical debt and redundant functionality.

The skill goes beyond simple file checking by implementing a sophisticated scoring system based on progressive disclosure and structural quality. It identifies orphan files, detects duplicate logic across different scopes, and provides a clear map of category coverage, ensuring your agent always has the right tools for the right task without the bloat of stale or poorly documented skills.

Skill-Audit Use Cases

  • Reviewing a local library of Openclaw Skills for structural issues and best-practice compliance.
  • Identifying duplicate or overlapping skill functionality across project and global directories.
  • Validating that documentation contains high-signal sections like Pitfalls or Common Issues to improve agent reliability.
  • Mapping category gaps to identify which areas of automation (e.g., CI/CD, Infrastructure) lack supporting skills.

How Skill-Audit Works

  1. The user triggers the audit by requesting a skill check or running the audit command.
  2. The agent executes the scripts/audit.sh utility to crawl all registered skill locations.
  3. Metadata is collected from YAML frontmatter and file structures to evaluate scoring criteria.
  4. Each skill is scored on a scale of 0-10 based on description quality, directory structure, and specific documentation sections.
  5. A comprehensive Markdown report is generated in the .sub-agent-results/ directory, detailing findings and prioritized fixes.

Skill-Audit Setup

To initialize the auditing environment for your Openclaw Skills, ensure the script is executable and trigger the audit via the agent interface:

# Ensure the audit script has execution permissions
chmod +x scripts/audit.sh

# Trigger the audit within your agent chat
"audit my skills"

Skill-Audit Data Schema & Taxonomy

The skill organizes audit data into a structured report with the following taxonomy:

Data Point Description
YAML Frontmatter Extracts name, description, and compatibility tags.
Scorecard Numeric evaluation (0-10) based on structural benchmarks.
Category Map Classification into one of 9 industry-standard skill categories.
Result Path Reports are saved to .sub-agent-results/skill-audit-report.md.

Skill-Audit Advanced Features

  • Progressive Disclosure Analysis: Checks for subdirectories like scripts/, references/, and assets/ to ensure complex skills are organized effectively.
  • Trigger-Oriented Linting: Analyzes the description field to verify it explicitly tells the model when to activate the skill.
  • Cross-Scope Deduplication: Detects functional overlaps between global, workspace, and project-specific skill directories.
  • Automated Scorecarding: Provides an instant impact-sorted list of issues to fix, categorized by severity (Good, Needs Work, Poor).

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*