Clawdex is a security auditing tool that verifies the safety of AI agent skills before they are installed to protect your system from malicious code.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install clawdex
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 clawdex 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 Clawdex by Koi?
Clawdex by Koi provides a critical security layer for developers and teams using agentic AI. By querying the Clawdex API, users can instantly verify if a specific skill from the ClawHub repository is safe, malicious, or pending audit. Powered by the Wings agentic AI risk engine, this tool ensures that Openclaw Skills do not contain harmful payloads such as backdoors, credential stealers, or unauthorized data exfiltration scripts, maintaining the integrity of your local development environment.
In an era where third-party scripts can pose significant risks, Clawdex offers peace of mind by providing a transparent verdict on the reliability of any skill. It bridges the gap between productivity and security, allowing users to leverage the power of Openclaw Skills without exposing their systems to malware or unauthorized resource usage.
Clawdex by Koi Use Cases
- Verifying the safety of a new skill found on ClawHub before running the installation command.
- Performing a security audit on a directory of previously installed Openclaw Skills to identify retroactive threats.
- Automating security checks within agent workflows to prevent the execution of unaudited or malicious code.
- Identifying high-risk behaviors like hidden crypto miners or reverse shells in third-party integrations.
How Clawdex by Koi Works
- The user identifies a specific Openclaw Skill they intend to install or audit.
- A request is sent to the Clawdex API endpoint using the skill identifier.
- The Wings AI risk engine analyzes the skill package and returns a JSON response containing a security verdict.
- If the verdict is benign, the installation proceeds; if malicious, the process is blocked to protect the system.
- For unknown verdicts, the system alerts the user to perform a manual review before granting approval.
Clawdex by Koi Setup
To verify a skill before installation, use the following curl command in your terminal:
curl -s "https://clawdex.koi.security/api/skill/SKILL_NAME"
To audit your existing Openclaw Skills, you can run a bulk check with this script:
# List installed skills and check each one
for skill in $(ls ~/.clawdbot/skills/); do
echo "Checking: $skill"
curl -s "https://clawdex.koi.security/api/skill/$skill"
echo
done
Clawdex by Koi Data Schema & Taxonomy
Clawdex provides a streamlined JSON response for easy integration into automated scripts and security workflows.
| Property | Value Type | Description |
|---|---|---|
| verdict | string | The security assessment: benign, malicious, or unknown. |
Verdict Definitions
- benign: The skill has been audited and no malicious behavior was found.
- malicious: The skill is harmful and should not be installed.
- unknown: The skill has not been audited yet and requires manual user approval.
Clawdex by Koi Advanced Features
- Real-time threat detection powered by Koi's Wings agentic AI risk engine for all Openclaw Skills.
- Comprehensive malware signatures covering credential theft, exfiltration, and reverse shells.
- Automated bulk auditing capabilities for maintaining security health across large local skill libraries.
- Seamless CLI integration for pre-installation safety checks and agent-led security enforcement.
SKILL.md
Loading
Related Openclaw Skills
MiniMax Usage Monitor
A streamlined CLI tool to monitor MiniMax Coding Plan prompt usage and subscription health in real-time.
QQ Bot Multi-Persona Management
A sophisticated management skill that allows QQ bots to switch between multiple personalities based on chat context, user ID, or channel type.
Tushare (Chinese Financial Data)
A specialized financial data interface for fetching comprehensive Chinese stock, futures, and macroeconomic datasets via the Tushare API.
System Design Interview Coach
A professional AI coach that guides you through the 8 stages of a system design interview using quantitative estimation.
Local Falcon AI Visibility & Local SEO Expert
An expert skill for optimizing local business visibility across traditional map packs and emerging AI search platforms like ChatGPT and Google AI Overviews.
RapidAPI Universal Skill
A template-driven RapidAPI client that converts API endpoints into stable, callable actions for AI agents.
Featured*
