# Vet OpenClaw Skills Before Installation

Installing unreviewed skills can expose projects to unsafe permissions, hidden instructions, or typosquat risk. This skill provides a structured checklist for manual pre-install security review.

## Install

```bash
npx skillstore add useai-pro/skill-vetter
```

## Metadata

- Status: approved
- Slug: useai-pro-skill-vetter
- Version: 1.0.0
- Author version: 1.0.0
- Skillstore revision: r1
- Version status: valid
- Tree hash: 6fb8d4473bf82e1b18ade9b0c347fcbf9bd08e256d7dc2e7a69c99973f561cf8
- Author: useai-pro
- GitHub username: useai-pro
- License: MIT
- Repository: https://github.com/useai-pro/openclaw-skills-security/tree/main/skills/skill-vetter/
- Ref: 26421118b848d9f1efc0aa169d8a7a9e7e0a877e
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: confirmation\_required
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands, filesystem
- Quality score: 38
- Quality tier: warning
- Public page: https://skillstore.pages.dev/skills/useai-pro-skill-vetter
- Manifest: https://skillstore.pages.dev/api/skills/useai-pro-skill-vetter/manifest

## Capabilities

- Reviews SKILL.md metadata for name, version, description, and author signals.
- Classifies requested permissions by risk and required justification.
- Lists critical red flags such as credential paths, obfuscation, unsafe commands, and sandbox bypass requests.
- Checks skill names for typosquat patterns and common misspellings.
- Produces a manual report with verdict, permissions, red flags, and installation recommendation.

## Use Cases

- Review a marketplace submission: Assess a submitted OpenClaw skill before allowing it into a shared catalog.
- Check a GitHub skill before install: Review a SKILL.md file from a repository before adding it to a local agent setup.
- Audit installed skills periodically: Recheck existing skills after version changes or permission updates.

## Prompt Templates

### Basic Skill Check

```
Vet this SKILL.md before I install it. Focus on metadata, permissions, red flags, and a clear install recommendation.
```

### Permission Review

```
Review this skill for permission risk. Explain whether each requested permission is necessary and what evidence supports the decision.
```

### Typosquat and Identity Review

```
Check this skill name, author, and source for typosquat or impersonation risk. Compare it against likely legitimate names.
```

### Full Pre-Install Audit

```
Run a full pre-install vetting pass on this skill. Include metadata, permissions, content red flags, typosquat risk, verdict, and remediation steps.
```

## Limitations

- It is a checklist skill, not an automated malware scanner.
- It depends on the reviewer providing the complete skill content.
- It cannot verify external author reputation without additional evidence.
- It does not execute or sandbox the reviewed skill.

## Best Practices

- Review the full SKILL.md file, including frontmatter and body text.
- Require clear justification for file write, network, shell, and elevated access.
- Treat version updates as new reviews when permissions or instructions change.

## Anti Patterns

- Installing a skill based only on popularity or a familiar name.
- Ignoring quoted instructions that ask the agent to bypass safety controls.
- Approving broad permissions without exact files, commands, endpoints, and reasons.

## Security Audit

- Audited at: 2026-07-09T14:07:14.067\+00:00
- Summary: The static command, filesystem, sensitive-file, and prompt-injection hits are false positives caused by Markdown examples and defensive checklist text. The skill contains no executable script, network call, or direct secret access request. One semantic issue remains: the community skill self-declares audit trust metadata that could mislead users without independent verification.

## Stats

- Views: 127
- Downloads: 61
- Favorites: 0
- Popularity score: 0