# Scan AI Skills Before Use

AI skills can contain risky scripts, network calls, or unsafe install behavior. SkillScan checks packages and installed skills through a cloud scanner before use.

## Install

```bash
npx skillstore add tokauth/skillscan
```

## Metadata

- Status: approved
- Slug: tokauth-skillscan
- Version: 1.1.5
- Author version: 1.1.5
- Skillstore revision: r1
- Version status: valid
- Tree hash: 799a9d6435d7cbd10e264112bf90ac8501cf5665eb331220fd29ce3716e4ed22
- Author: tokauth
- GitHub username: tokauth
- License: MIT
- Repository: https://github.com/tokauth/SkillScan/tree/main/
- Ref: 36e07d5e13068e5be64447e8f20b427cf2cbd21a
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: blocked
- Manual install advisory: allowed\_with\_warning
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: network, filesystem, env\_access, external\_commands
- Quality score: 38
- Quality tier: warning
- Public page: https://skillstore.pages.dev/skills/tokauth-skillscan
- Manifest: https://skillstore.pages.dev/api/skills/tokauth-skillscan/manifest

## Capabilities

- Scans a single skill directory or zip archive for security risk.
- Finds installed skill directories across common AI agent locations.
- Calculates directory SHA-256 hashes for cloud cache lookup.
- Uploads packaged skill content to a cloud API when no cache result exists.
- Displays SAFE, LOW, MEDIUM, HIGH, or CRITICAL verdicts with threat labels.
- Prompts before deleting installed skills that receive high-risk verdicts.

## Use Cases

- Review a New Skill Package: Scan a downloaded zip or directory before installing it into an AI agent environment.
- Audit Existing Agent Skills: Find installed skills across local agent folders and scan them as a batch.
- Gate Marketplace Submissions: Use scan verdicts and threat labels as an input before approving community skills.

## Prompt Templates

### Scan One Skill

```
Scan this skill directory for security risk and summarize the verdict before I use it: <path>
```

### Scan Installed Skills

```
Find my installed AI skills, scan them, and list any MEDIUM, HIGH, or CRITICAL results.
```

### Assess a Shared Archive

```
Scan this skill zip archive and explain whether it is acceptable to install: <path-to-zip>
```

### Prepare Remediation Guidance

```
Use the scan findings to identify the risky behaviors and propose concrete fixes for the skill author.
```

## Limitations

- Uncached scans require network access to the SkillScan cloud service.
- Skill contents may be uploaded to an external service for analysis.
- The auto-update path trusts a remote manifest and does not use independent signatures.
- Scan accuracy depends on the remote API response and service availability.

## Best Practices

- Scan skills before installation, update, or first use.
- Review network and filesystem findings before trusting a cloud scan result.
- Keep copies of scan reports outside the skill directory.

## Anti Patterns

- Do not treat a SAFE result as proof that all future updates are safe.
- Do not upload private or proprietary skill code without policy approval.
- Do not enable silent self-updates without user consent and signature verification.

## Security Audit

- Audited at: 2026-07-07T07:42:31.578\+00:00
- Summary: SkillScan performs real security-scanning work, but it relies on external cloud requests, uploads skill archives, collects host fingerprint data, and silently self-updates. Many static hits are Markdown formatting or guarded temporary-file use, but the remote update, persistent instruction, and data-transfer behaviors need remediation before publication.

## Stats

- Views: 109
- Downloads: 16
- Favorites: 2
- Popularity score: 0