Setup & Installation
openclaw skills install @minduploadedcrab/minduploadedcrab-skillguard
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
npx clawhub install minduploadedcrab-skillguard
What This Skill Does
SkillGuard scans OpenClaw skill directories for security threats before installation. It detects credential theft, data exfiltration, prompt injection, obfuscation, and permission overreach. Produces a risk score and verdict for each scanned skill.
Generic antivirus misses agent-specific attack vectors like prompt injection and symlink escapes that SkillGuard is built to catch.
When to use it
- Scanning a third-party skill before installing it
- Auditing all currently installed skills for backdoors
- Checking a skill for hardcoded API keys or secrets
- Detecting prompt injection hidden in a skill's SKILL.md
- Reviewing dependency scripts for malicious post-install hooks
Example Workflow
Here's how your AI assistant might use this skill in practice.
INPUT
User asks: scan the skill at ~/.openclaw/workspace/skills/some-skill before I install it
AGENT
- 1Runs `python3 scripts/skillguard.py scan ~/.openclaw/workspace/skills/some-skill --json`
- 2Parses the JSON output for risk score, verdict, and findings
- 3Checks for FAIL or WARN verdicts and surfaces specific findings with severity
- 4Reports credential access or network exfiltration detections if present
- 5Presents the final verdict and recommends whether to proceed with installation
OUTPUT
Risk Score: 42/100, Verdict: WARN — one finding: outbound HTTP call to unrecognized domain in index.js line 14