A local-first security scanner designed to detect malicious patterns and vulnerabilities in AI agent skills before they are installed.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install moltcops
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 moltcops 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 MoltCops?
MoltCops is a dedicated security auditing tool built to protect the ecosystem of Openclaw Skills from emerging threats. It provides a critical pre-installation check by scanning skill directories for 20 distinct threat categories, including prompt injection, data exfiltration, and unauthorized lateral movement. Unlike traditional signature-based malware scanners, it focuses on behavioral patterns that could compromise an agent's integrity or leak sensitive user data.
Designed with a local-first philosophy, MoltCops ensures that your source code and configurations never leave your machine. It requires no external API calls, accounts, or dependencies, making it a lightweight yet powerful gatekeeper for any developer working with autonomous agents and third-party codebases.
MoltCops Use Cases
- Scanning third-party Openclaw Skills from ClawHub or GitHub before deployment.
- Auditing code shared by other agents to prevent prompt injection or tool-use steering.
- Validating unknown scripts for credential harvesting or financial drain patterns.
- Integrating security gates into CI/CD pipelines for automated agent skill verification.
How MoltCops Works
- The scanner is pointed to a local directory containing the source code of the skill.
- It executes a comprehensive suite of 20 detection rules using the Python standard library.
- The engine analyzes patterns for shell injections, hardcoded secrets, and destructive force flags.
- It applies context-aware filtering to reduce false positives for legitimate environment variable or git operations.
- The tool generates a structured report with a verdict of PASS, WARN, or BLOCK based on the severity of the findings.
MoltCops Setup
MoltCops is designed to run with zero external dependencies, requiring only Python 3. To install and run a scan on a specific skill, use the following commands:
# Navigate to the MoltCops directory and scan a target folder
python3 scripts/scan.py <path-to-skill-folder>
# Example: Scanning a suspicious skill in your local directory
python3 scripts/scan.py ./suspicious-skill
MoltCops Data Schema & Taxonomy
MoltCops organizes its security findings into a structured taxonomy. When scanning Openclaw Skills, the results are categorized as follows:
| Category | Rule IDs | Example Patterns |
|---|---|---|
| Prompt Injection | MC-001 - MC-003 | System prompt overrides, tool-use steering |
| Code Injection | MC-004 - MC-006 | Shell injections, eval/exec, base64 payloads |
| Data Exfiltration | MC-007 - MC-010 | Webhook URLs, credential harvesting |
| Hardcoded Secrets | MC-011 - MC-012 | API keys, private key material |
| Autonomy Abuse | MC-017 - MC-018 | Destructive flags (rm -rf), sudo escalation |
MoltCops Advanced Features
- Local-first architecture ensures data privacy with no external uploads or API dependencies.
- Context-aware filtering skips standard git remotes and only flags high-risk environment variables.
- Triple-verdict exit codes (0, 1, 2) allow for easy integration into automated DevOps workflows for Openclaw Skills.
- Behavioral detection catches sleeper triggers and drain logic that standard antivirus software often misses.
SKILL.md
Loading
Related Openclaw Skills
MoltCops
A local-first security scanner designed to detect malicious patterns and vulnerabilities in AI agent skills before installation.
Migma
A terminal-based AI email engine for designing, sending, and managing marketing campaigns and contact segments.
OpenAI Codex CLI
A powerful terminal-based coding assistant that integrates OpenAI Codex directly into your local development workflow for automated refactoring, debugging, and feature implementation.
safe-web: Secure Web Search and Fetching
A security-focused web fetching and search tool that shields AI models from malicious prompt injections found in live web content.
Python Coding Guidelines
A comprehensive set of Python coding standards and automated checks designed to ensure high-quality, idiomatic, and maintainable code.
Competitor Finder Skill
An automated market intelligence tool that identifies top competitors using a three-tier fallback strategy including search APIs and AI models.
METADATA
Featured*
