A deep auditing and optimization CLI utility for local Openclaw Skills environments, providing comprehensive usage analysis, permission reviews, and automated cleanup scripts.
The fastest way to install a skill directly from the registry.
npx clawhub@latest install audit-tool
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
Copy this prompt to OpenClaw to install it automatically.
Help me install audit-tool using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Get the raw skill files in a ZIP archive.
Skill Auditor is a robust diagnostic utility built specifically for managing local agent environments. It scans local installations to read metadata, files, permissions, dependencies, and execution history. As your library of Openclaw Skills grows, Skill Auditor ensures your agent system remains lean, performant, and secure by detecting over-privileged actions, finding duplicate functionality, and pruning inactive skills.
By assessing five core vectors—Activity, Security, Stability, Maintainability, and Documentation quality—this tool delivers a comprehensive health score from 0-100. It empowers developers and system administrators to maintain clear oversight of AI agent behavior and resource utilization, generating ready-to-execute cleanup scripts to optimize local environments safely.
~/.openclaw/skills/ to read skill configurations, dependencies, licenses, and permission declarations.To set up the skill-auditor, make sure it is installed inside your local agent workspace.
# Install the skill auditor within your openclaw workspace
openclaw install skill-auditor
# Run a baseline audit to check existing installations
skill-auditor audit
Configure the default tracking window by passing parameters during CLI execution:
# Scan skills with customized parameters (e.g., look back 90 days)
skill-auditor audit --days 90 --min-activity 3
# Run cleanup tasks in dry-run mode to inspect proposed changes
skill-auditor cleanup --dry-run
The utility generates structured reports and parses key configurations across several metadata dimensions. Below is the framework used to score and organize Openclaw Skills data:
| Dimension | Target Area | Description | Weight |
|---|---|---|---|
| Activity | Usage patterns | Analyzes total invocations and recent activity to flag zombie skills. | 20% |
| Security | Access permissions | Flags over-privileged tools with high-risk access (exec, web_fetch, etc.). |
20% |
| Stability | System errors | Evaluates completeness and historically logged runtime errors. | 20% |
| Maintainability | Structural health | Scans overall file counts, bundle sizes, and namespace conflicts. | 20% |
| Docs | Verification | Verifies description clarity, sample prompts, and overall metadata completeness. | 20% |
All generated cleanup operations output a shell script containing structured commands (uninstall, disable, suspend) mapping directly to verified skill IDs.
--focus security).Loading
An intelligent travel agent that generates optimized day-by-day itineraries, custom budgets, transport routing, and contingency plans based on user preferences.

An AI-powered nutrition analysis tool equipped with a 2,000+ entry Chinese food database, designed to parse natural language meal descriptions and track goal-oriented macro targets.

An automation workflow that guides developers through implementing and verifying a dynamic account dropdown filter based on the chosen payment method.

An AI-powered knowledge base orchestrator that automates data ingestion, querying, linting, and synthesis for Obsidian wikis.

A powerful utility to extract, structure, and link decisions, action items, and knowledge from raw AI chat sessions into Markdown, JSON, or Obsidian.

An automated git safety auditing agent that flags risk signals, maps impact domains, and outputs a structured risk score from git blame logs.








































