A diagnostic utility for scanning AI agent workspaces to detect similar or duplicate skills that may cause model confusion.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install skill-dedup-scanner
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install skill-dedup-scanner 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 Dedup Scanner?
The Skill Dedup Scanner is a professional-grade audit tool designed to maintain the integrity of an AI agent's workspace. By analyzing the names and descriptions of installed Openclaw Skills, the scanner identifies potential conflicts that could lead to incorrect tool selection by Large Language Models. It provides a structured approach to identifying overlapping functionality, ensuring that your agent always has a clear and unique set of capabilities to choose from.
Built with developer workflows in mind, this tool supports multi-language environments and provides detailed similarity reports. Whether you are managing a local collection or preparing to publish to a registry, this scanner is a critical piece of the development lifecycle for anyone building high-quality Openclaw Skills.
Skill Dedup Scanner Use Cases
- Auditing a large workspace to find redundant or overlapping tool definitions.
- Troubleshooting an AI agent that is triggering the wrong skills during execution.
- Validating new skills before publishing them to ensure they don't conflict with existing ones.
- Standardizing naming conventions across a multi-developer team environment.
How Skill Dedup Scanner Works
- The scanner traverses the specified directory to locate all markdown files containing skill definitions.
- It parses the YAML frontmatter to extract the core metadata, including the skill name and description.
- The similarity engine compares every skill against the rest of the collection using a customizable threshold algorithm.
- It categorizes conflicts into High, Medium, or Safe levels based on calculated similarity scores.
- The tool generates a report (in Markdown or JSON) with actionable recommendations for renaming or merging skills.
Skill Dedup Scanner Setup
Ensure you have Python 3.7 or higher installed on your system. First, install the necessary dependencies:
pip install pyyaml
To perform a basic scan of your local environment, run the following command:
python3 scripts/main.py ~/.openclaw/workspace/skills/
Skill Dedup Scanner Data Schema & Taxonomy
The Skill Dedup Scanner organizes its findings through a structured reporting system and a multi-language locale architecture.
| Component | Description |
|---|---|
| Frontmatter | Parses name and description from skill markdown files. |
| Thresholds | Uses a 0.0 to 1.0 scale to determine similarity (default 0.7). |
| Locales | Uses JSON files in the /locales folder for EN and ZH translations. |
| Output | Can export reports as standard output, Markdown files, or structured JSON. |
Skill Dedup Scanner Advanced Features
- Customizable similarity thresholds via the
--thresholdflag for granular control over conflict detection. - Full JSON output support for integration into automated CI/CD pipelines and developer tools.
- Automatic language detection for English and Chinese, ensuring global compatibility for Openclaw Skills developers.
- Verbose reporting mode for deep-dive analysis into specific name and description similarity percentages.
SKILL.md
Loading
Related Openclaw Skills
Video URL to Text (Toncall)
An automated pipeline to download video, extract audio, and generate transcriptions using Volcengine ASR for Openclaw Skills users.
Fontpreview
A fast, local-first documentation utility for Fontpreview patterns and developer best practices.
Mathbox
A local, high-speed reference tool providing immediate access to Mathbox implementation patterns, debugging guides, and performance best practices.
Project Context Manager
A standardized system for managing the lifecycle of long-term projects, ensuring reliable context recovery and cross-session persistence.
Frontend CV
A sophisticated resume engine that transforms raw career data into high-quality, print-ready HTML and PDF portfolios.
genai-calling
A unified interface and SDK for interacting with multiple AI providers across all modalities including text, image, audio, video, and embeddings.
Featured*
