Skill Guard v5.0

Install: clawhub install tommo-skill-guard

Security scanning for OpenClaw skills. Zero exec — read-only scanning via the built-in read tool.

Language

Detect from user's message language. Default: English.

Pre-Install Check

When user wants to install a skill, check BEFORE installing:

1. Navigate to https://clawhub.ai/skills/{slug} via browser
2. Snapshot and look for Security Scan section
3. Report findings:

If browser unavailable: clawhub inspect {slug} for basic metadata.

Local Pattern Scan

Scan installed skill files for dangerous patterns using the read tool only — no exec, no shell, no injection risk.

1. read ./skills/{name}/SKILL.md
2. List additional files with read if scripts/ or references/ exist
3. Search for patterns in the content:

Report format:

Scan: {skill-name}
  Files checked: {N}
  🔴 [file:line] {pattern} — {risk description}
  ✅ No issues found

Integrity Check

Compare files by reading them and noting their content fingerprint (first/last lines + file size). No hashing exec needed — the read tool is sufficient for detecting file changes.

Baseline (user-initiated only):

• User says "baseline {skill}"
• Agent reads all files in ./skills/{name}/
• Saves file list + sizes + first/last lines to memory/skill-guard/{name}-baseline.txt
• Shows the baseline to user for review

Verify (user-initiated only):

• User says "integrity check {skill}"
• Agent reads current files and compares against saved baseline
• Reports any differences

Auto-baseline is disabled by design. New skills are never automatically trusted.

Quick Commands

Guidelines for Agent

1. Use read only — never exec, never shell, no command injection possible
2. Validate skill names — only scan skills in ./skills/ directory
3. Never auto-baseline — user must explicitly request
4. Always show findings — never silently block or allow
5. User decides — show risk, let user choose

What This Skill Does NOT Do

• Does NOT use exec, shell, or any subprocess execution
• Does NOT auto-baseline newly installed skills
• Does NOT block installations automatically
• Does NOT modify skill files
• Does NOT require Node, bash, curl, or any external tool
• Does NOT access credentials or private data
• Does NOT write files outside memory/skill-guard/ (explicit user request only)

More by TommoT2

• setup-doctor — Diagnose and fix OpenClaw setup issues
• context-brief — Persistent context survival across sessions
• skill-analytics — Monitor skill portfolio performance

Install the full suite:

clawhub install tommo-skill-guard setup-doctor context-brief skill-analytics