Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
Version History
- Removed always-on flag (`always: true`) from skill metadata, making the scanner no longer run by default in all cases. - Updated documentation to clarify and simplify examples of high-risk patterns, replacing real domains and paths with generic examples. - No changes to core functionality; update is documentation and configuration only.
What This Skill Does
Scans OpenClaw skills for security issues, suspicious code patterns, and excessive permissions. Produces a trust score from 0 to 100 based on author reputation, permission scope, and code analysis to help users decide whether a skill is safe to install or run.
Structured trust scoring with weighted criteria replaces manual code review, making security decisions faster and more consistent across skills.
When to use it
- Checking a new ClawHub skill before installing it
- Running a weekly audit of all installed skills
- Getting a quick trust score for an unfamiliar skill
- Investigating a skill after a security incident in the ecosystem
- Reviewing permissions requested by a skill before execution
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Is this skill from ClawHub safe to install?
- 1Runs scan-skill on the skill's local path
- 2Analyzes frontmatter metadata for requested bins and environment variables
- 3Checks SKILL.md content for suspicious patterns like network exfiltration, credential access, or obfuscated commands
- 4Calculates trust score weighted by author reputation, permission scope, code patterns, update frequency, and download count
- 5Returns a formatted scan result with issues by severity and a recommendation
Trust score report showing score out of 100, list of permissions requested, flagged issues with severity levels, and a clear install or skip recommendation