Before you install
Confirm what the skill asks the agent to read, write, execute, download, or send outside the environment.
OpenClaw skills
Check an OpenClaw skill before installation. AI Skill Shield helps you inspect SKILL.md instructions, supporting files, and repository signals that can affect your agent and its environment.
Scan an OpenClaw skillConfirm what the skill asks the agent to read, write, execute, download, or send outside the environment.
A safe-looking skill file is not enough when scripts, dependencies, workflows, and configuration can alter its behavior.
Compare requested access with the skill’s stated job and flag permissions that are unnecessary or unusually broad.
Use the report to document why a skill was approved, restricted, or rejected before it reaches production workflows.
Paste the GitHub repository URL, upload the skill package, or provide the SKILL.md content. Review the findings and installation recommendation before giving the skill access to a real environment.