skill-safe-install

What This Skill Does

The skill-safe-install is a critical L0-level security framework designed for the OpenClaw agent ecosystem. Its primary function is to enforce a rigorous, six-step verification pipeline whenever a user initiates the installation of a new skill. By requiring systematic checks—including duplication analysis, security inspection of dependencies, sandboxed testing, and manual authorization for whitelist entries—it protects your environment from unauthorized code execution, dependency conflicts, and malicious persistence attempts.

Installation

To integrate this safety protocol into your OpenClaw environment, execute the following command: clawhub install openclaw/skills/skills/halfmoon82/skill-safe-install-l0 Ensure that your OpenClaw runtime has appropriate permissions to modify the central configuration file and execute system-level commands, as this skill interacts directly with clawhub binaries.

Use Cases

This skill is indispensable for:

• Enterprise Environments: Where every installed plugin must be audited for compliance.
• High-Security Setups: Preventing 'shadow IT' where unauthorized skills might leak API keys or modify system files.
• Developer Workflows: Providing a standardized, safe environment for testing third-party contributions before promoting them to the core allow-list.

Example Prompts

1. "Install the browser-automation skill using the safe installation procedure."
2. "I need to add the data-analysis module; please perform a security check first."
3. "Security check required: install the system-monitor skill."

Tips & Limitations

• Strict Workflow: The process is designed to be blocking. You cannot bypass steps; if an inspection (Step 2) fails, the installation is automatically aborted to protect your system.
• Sandbox Reality: Because the current clawhub lacks a native --sandbox flag, this skill uses a temporary directory via mktemp. Always ensure your system has sufficient disk space for these temporary workdirs.
• Manual Auth: Step 5 requires a user confirmation to modify the allowBundled whitelist. Do not automate this confirmation unless you have fully vetted the source code of the target skill.