Skip to content
Back to search

claude-skill-antivirus

Security scanner for Claude Code Skills — 9 engines detect malicious patterns, data exfiltration, dangerous ops across 71K+ skills

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/claude-world--claude-skill-antivirus/badge.svg)](/listings/claude-world--claude-skill-antivirus)

HTML

<a href="/listings/claude-world--claude-skill-antivirus"><img src="/api/listings/claude-world--claude-skill-antivirus/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

npm install -g claude-skill-antivirus

Or run directly with npx:

npx claude-skill-antivirus <skill-source>

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 448 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. # Claude Skill Antivirus
  2. [![npm version](https://img.shields.io/npm/v/claude-skill-antivirus.svg)](https://www.npmjs.com/package/claude-skill-antivirus)
  3. [![CI](https://github.com/claude-world/claude-skill-antivirus/actions/workflows/ci.yml/badge.svg)](https://github.com/claude-world/claude-skill-antivirus/actions/workflows/ci.yml)
  4. [![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
  5. [![Node.js](https://img.shields.io/badge/node-%3E%3D22.0.0-brightgreen.svg)](https://nodejs.org/)
  6. [![Claude Code](https://img.shields.io/badge/Claude_Code-Opus_5.5-blueviolet.svg)](https://docs.anthropic.com/en/docs/claude-code)
  7. <img src="assets/demo.gif" alt="Two scans in a terminal: a safe skill scores LOW 90/100, then a skill hiding code in scripts/ scores CRITICAL 0/100 with five critical findings" title="A clean skill passes; one hiding code in scripts/ is caught" />
  8. A security scanner and safe installer for Claude Code Skills. It scans the **whole skill directory** — not just `SKILL.md` — to detect malicious patterns, hidden code, data exfiltration attempts, and dangerous operations before installing third-party skills.
  9. Compatible with Claude Code using Opus 5.5, Sonnet 5.5, Haiku 5.5, and Fable 5.1 models.
  10. [繁體中文說明](./README.zh-TW.md) | [SkillsMP Scan Report](./SCAN-REPORT.md)
  11. ## SkillsMP Platform Scan Results
  12. We scanned all **71,577 skills** on SkillsMP:
  13. | Risk Level | Count | Percentage |
  14. |------------|-------|------------|
  15. | CRITICAL | 91 | 0.13% |
  16. | HIGH | 626 | 0.87% |
  17. | MEDIUM | 1,310 | 1.83% |
  18. | SAFE | **69,505** | **97.11%** |
  19. **~3% of skills may have potential risks.** See [full report](./SCAN-REPORT.md) for details.
  20. > **Coverage note**: this catalog scan (v2.0.0, January 2026) used the SkillsMP API, which exposes only each skill's `SKILL.md`, so "SAFE" there means no findings in the manifest; scripts were not analyzed (see [#33](https://github.com/claude-world/claude-skill-antivirus/issues/33)). Scan a skill's directory or repository for a complete verdict.
  21. > **Note**: Some findings may be false positives (e.g., legitimate 1Password/Bitwarden integrations). Manual review is recommended for flagged skills.
  22. ## Features
  23. - **Whole-directory scanning**: every file in a skill is analyzed — `scripts/`, nested
  24. folders, and even files inside `node_modules/`, `venv/`, or `__pycache__/`. Hidden code
  25. can no longer evade the scan by living outside `SKILL.md` (see [#33](https://github.com/claude-world/claude-skill-antivirus/issues/33)).
  26. - **10 Security Scanning Engines**:
  27. - Dangerous Commands Scanner - Detects destructive shell commands
  28. - Data Exfiltration Scanner - Identifies data theft patterns
  29. - External Connections Scanner - Analyzes URLs and network calls
  30. - Permission Scanner - Reviews tool permissions and access scope
  31. - Pattern Scanner - Detects prompt injection and sensitive data
  32. - MCP Security Scanner - Validates MCP server configurations
  33. - SSRF Scanner - Identifies server-side request forgery patterns
  34. - Dependency Scanner - Detects malicious packages and typosquatting
  35. - Sub-agent Scanner - Detects Task tool abuse and agent chain attacks
  36. - File Artifact Scanner - Flags compiled/binary/archive artifacts (`.pyc`, `.so`, `.exe`, archives)
  37. - **Tamper-evident coverage**: symlinks are never followed out of the skill, and skipped
  38. directories, truncated files, and scan-limit hits are reported as findings so an unusual
  39. input can never silently pass as SAFE.
  40. - **Risk Assessment**: Critical, High, Medium, Low, and Info levels
  41. - **Multilingual Support**: English and Traditional Chinese (繁體中文)
  42. - **Install or Scan-Only Mode**: Review skills before installation
  43. - **Interactive Prompts**: Guided decision-making for risky installations
  44. ## Requirements
  45. - Node.js >= 22
  46. ## Installation
  47. ```bash
  48. npm install -g claude-skill-antivirus
  49. ```
  50. Or run it without installing:
  51. ```bash
  52. npx claude-skill-antivirus <skill-source>
  53. npx claude-skill-antivirus ./path/to/skill --scan-only
  54. ```
  55. ## Usage
  56. ### Install a skill with security scanning
  57. ```bash
  58. # Install to project level (./.claude/skills/) - default
  59. skill-install ./path/to/skill
  60. skill-install https://github.com/user/skill-repo
  61. # Install to user level (~/.claude/skills/)
  62. skill-install ./path/to/skill --global
  63. skill-install @skillsmp/example-skill -g
  64. ```
  65. **Installation paths:**
  66. - Project level (default): `./.claude/skills/`
  67. - User level (`--global`): `~/.claude/skills/`
  68. ### Scan only (without installing)
  69. ```bash
  70. skill-install ./path/to/skill --scan-only
  71. skill-install https://github.com/user/repo/tree/main/skills/my-skill --scan-only
  72. ```
  73. Every file in the skill is scanned. GitHub URLs (repository, `tree/` directory, or a `SKILL.md` blob) are walked through the GitHub API; set `GITHUB_TOKEN` to avoid unauthenticated rate limits. A skill that could not be scanned completely is reported, and is never installed.
  74. ### Change language
  75. ```bash
  76. # English (default)
  77. skill-install ./path/to/skill --lang en
  78. # Traditional Chinese
  79. skill-install ./path/to/skill --lang zh-TW
  80. ```
  81. ### Command names
  82. A global install provides three interchangeable names for the installer, so
  83. every example above works with any of them:
  84. ```bash
  85. claude-skill-antivirus ./path/to/skill --scan-only
  86. skill-install ./path/to/skill --scan-only
  87. claude-skill-av ./path/to/skill --scan-only
  88. ```
  89. `skill-batch-scan` is the separate batch tool described next.
  90. ### Batch scan all SkillsMP skills
  91. ```bash
  92. # Scan all skills from SkillsMP (requires API key)
  93. skill-batch-scan --api-key <your-api-key>
  94. # Scan with options
  95. skill-batch-scan --api-key <key> --max-pages 10 --verbose
  96. skill-batch-scan --api-key <key> --output ./my-reports --lang zh-TW
  97. ```
  98. Options:
  99. - `-k, --api-key <key>` - SkillsMP API key (required)
  100. - `-l, --limit <number>` - Skills per page (default: 100)
  101. - `-p, --max-pages <number>` - Maximum pages to scan (default: all)
  102. - `-o, --output <dir>` - Output directory for reports (default: ./scan-reports)
  103. - `-v, --verbose` - Show verbose output
  104. - `--lang <lang>` - Language (en, zh-TW)
  105. ## Scanning Engines
  106. ### 1. Dangerous Commands Scanner
  107. Detects commands that can cause system damage:
  108. | Risk Level | Detection Items |
  109. |------------|-----------------|
  110. | Critical | `rm -rf /`, `curl \| bash`, fork bombs |
  111. | High | Reading `/etc/shadow`, reverse shells, credential theft |
  112. | Medium | `rm -rf`, permission changes, service control |
  113. | Low | `sudo`, global package installs |
  114. ### 2. Permission Scanner
  115. Analyzes `allowed-tools` declarations:
  116. - **Critical**: `Bash(*)` - Unrestricted shell access
  117. - **High**: `Write`, `WebFetch`, broad bash permissions
  118. - **Medium**: `Read`, `Glob`, `Grep`, version control tools
  119. - **Dangerous Combinations**: e.g., `Read + WebFetch` = data exfiltration risk
  120. ### 3. External Connections Scanner
  121. Identifies suspicious network activity:
  122. - Direct IP URLs
  123. - Webhook/data capture services
  124. - Suspicious TLDs (.tk, .ml, etc.)
  125. - Discord/Telegram webhooks
  126. - URL shortening services
  127. ### 4. Pattern Scanner
  128. Detects:
  129. - Prompt injection attacks
  130. - Hardcoded credentials/API keys
  131. - Obfuscated code (base64, hex encoding)
  132. - Social engineering language
  133. ### 5. Data Exfiltration Scanner
  134. Specifically detects malicious behavior of reading local data and sending it externally:
  135. | Category | Detection Items |
  136. |----------|-----------------|
  137. | Data Collection | Reading `.ssh`, `.aws`, `.env`, browser passwords, password managers |
  138. | Data Exfiltration | `curl -d`, netcat transfers, DNS tunneling, email exfiltration |
  139. | Combined Attacks | `cat \| base64 \| curl`, `tar \| nc`, `find -exec curl` |
  140. | Env Variable Theft | `env \| curl`, `printenv` exfiltration |
  141. | System Recon | `whoami`, `hostname`, network config exfiltration |
  142. | Persistence | Modifying `.bashrc`, scheduled cron exfiltration |
  143. ### 6. MCP Security Scanner
  144. Detects security risks in MCP Server configurations:
  145. | Category | Detection Items |
  146. |----------|-----------------|
  147. | Untrusted Sources | Non-official MCP servers, direct URL execution |
  148. | Dangerous Permissions | Unrestricted filesystem access, shell execution, database access |
  149. | Sensitive Config | Environment variables with credentials, exposed config |
  150. | Dangerous Combinations | Filesystem + Fetch, Shell + Network |
  151. ### 7. SSRF Scanner
  152. Detects Server-Side Request Forgery and cloud attacks:
  153. | Category | Detection Items |
  154. |----------|-----------------|
  155. | Cloud Metadata | AWS/GCP/Azure 169.254.169.254, IAM credential theft |
  156. | Internal Network | 10.x.x.x, 192.168.x.x, 172.16-31.x.x probing |
  157. | SSRF Bypass | Hex IP, URL encoding, file://, gopher:// |
  158. | Kubernetes | API access, secrets theft, serviceaccount |
  159. | Docker | docker.sock access, privileged containers, container escape |
  160. ### 8. Dependency Scanner
  161. Detects malicious or vulnerable dependencies:
  162. | Category | Detection Items |
  163. |----------|-----------------|
  164. | Known Malicious | event-stream, ua-parser-js, colors, faker |
  165. | Typosquatting | crossenv, lodash-, mongose, reqeusts |
  166. | Suspicious Install | URL installs, insecure registry, HTTP index |
  167. | postinstall Risks | Install scripts with curl, wget, eval |
  168. ### 9. Sub-agent Scanner
  169. Detects Task tool and sub-agent abuse:
  170. | Category | Detection Items |
  171. |----------|-----------------|
  172. | Privilege Escalation | Task spawning Bash agent, requesting all permissions |
  173. | Prompt Injection | Sub-agent prompts with malicious commands |
  174. | Agent Chain Attacks | Nested Task calls, recursive agents |
  175. | DoS Attacks | Loop Task calls, infinite recursion |
  176. | Data Theft | Read + WebFetch combinations, accessing sensitive data |
  177. ### 10. File Artifact Scanner
  178. Flags non-text artifacts that cannot be reviewed as source and are a common hiding place for payloads:
  179. | Category | Detection Items |
  180. |----------|-----------------|
  181. | Compiled / native | `.pyc`, `.pyo`, `.so`, `.node`, `.dll`, `.dylib`, `.exe`, `.wasm`, `.class` |
  182. | Archives | `.zip`, `.tar`, `.gz`, `.7z`, `.jar`, `.whl` bundled inside a skill |
  183. | Binary content | Non-text files detected by content, with only their readable strings scanned |
  184. | Coverage anomalies | Unexpected directories, unfollowed symlinks, truncated files, scan-limit hits |
  185. ## Output Examples
  186. ### Safe Skill
  187. ```
  188. 🔧 Claude Skill Installer v2.2.1
  189. 📦 Skill loaded: example-safe-skill
  190. 🔍 Starting security scan...
  191. ===========================================
  192. SECURITY SCAN REPORT
  193. ===========================================
  194. Risk Level: ✅ SAFE
  195. 📊 Findings Summary:
  196. 🟢 CRITICAL: 0
  197. 🟢 HIGH: 0
  198. 🟢 MEDIUM: 0
  199. 🟢 LOW: 0
  200. ℹ️ INFO: 2
  201. ✅ Recommendation: Safe to install
  202. ```
  203. ### Malicious Skill Detected
  204. ```
  205. 🔧 Claude Skill Installer v2.2.1
  206. 📦 Skill loaded: suspicious-skill
  207. 🔍 Starting security scan...
  208. ===========================================
  209. SECURITY SCAN REPORT
  210. ===========================================
  211. Risk Level: ☠️ CRITICAL
  212. 📊 Findings Summary:
  213. 🔴 CRITICAL: 5
  214. 🟠 HIGH: 3
  215. 🟡 MEDIUM: 2
  216. 🟢 LOW: 1
  217. ℹ️ INFO: 4
  218. 🔴 CRITICAL Findings:
  219. • [Data Collection] Reading sensitive credential files
  220. Attempts to read environment variables, private keys or credential files
  221. • [Data Exfiltration] curl sending command output
  222. Using curl to send command execution results to external server
  223. ...
  224. ❌ Recommendation: DO NOT INSTALL - Contains critical security risks
  225. ```
  226. ## Risk Levels
  227. | Level | Score Impact | Action |
  228. |-------|--------------|--------|
  229. | CRITICAL | -30/item | Block installation |
  230. | HIGH | -20/item | Require explicit confirmation |
  231. | MEDIUM | -10/item | Show warning |
  232. | LOW | -5/item | Show in verbose mode |
  233. | INFO | 0 | Always show |
  234. ## API
  235. You can also use the scanner programmatically:
  236. ```javascript
  237. import { SecurityScanner, loadSkill } from 'claude-skill-antivirus';
  238. const scanner = new SecurityScanner();
  239. const skill = await loadSkill('./path/to/skill'); // local dir, SKILL.md, zip, or GitHub URL
  240. const result = await scanner.scan(skill);
  241. console.log(result.riskLevel); // 'SAFE' | 'LOW' | 'MEDIUM' | 'HIGH' | 'CRITICAL'
  242. console.log(result.score); // 0-100
  243. console.log(result.findings);
  244. // {
  245. // critical: [...],
  246. // high: [...],
  247. // medium: [...],
  248. // low: [...],
  249. // info: [...]
  250. // }
  251. ```
  252. ## Project Structure
  253. ```
  254. claude-skill-antivirus/
  255. ├── src/
  256. │ ├── index.js # CLI entry point
  257. │ ├── i18n/ # Internationalization
  258. │ │ ├── index.js
  259. │ │ ├── en.js # English translations
  260. │ │ └── zh-TW.js # Traditional Chinese translations
  261. │ ├── lib.js # Programmatic API entry (no CLI side effects)
  262. │ ├── version.js # Single source of truth for the version
  263. │ ├── scanner/
  264. │ │ ├── index.js # Main scanner (integrates 10 engines)
  265. │ │ ├── dangerous-commands.js # Dangerous command detection
  266. │ │ ├── permissions.js # Permission checking
  267. │ │ ├── external-connections.js # External connection analysis
  268. │ │ ├── patterns.js # Pattern matching
  269. │ │ ├── data-exfiltration.js # Data exfiltration detection
  270. │ │ ├── mcp-security.js # MCP Server security check
  271. │ │ ├── ssrf-scanner.js # SSRF/cloud attack detection
  272. │ │ ├── dependency-scanner.js # Dependency security check
  273. │ │ ├── subagent-scanner.js # Sub-agent attack detection
  274. │ │ └── file-artifacts.js # Compiled/binary/archive artifact detection
  275. │ └── utils/
  276. │ ├── downloader.js # Skill loader (walks the whole directory)
  277. │ └── installer.js # Skill installer
  278. ├── examples/
  279. │ ├── safe-skill/ # Safe example
  280. │ ├── malicious-skill/ # Malicious example (tests all engines)
  281. │ └── hidden-code-skill/ # Hidden-code example (payload outside SKILL.md, #33)
  282. ├── test/
  283. │ ├── scan.test.js # Regression suite (node --test)
  284. │ └── bin.test.js # Every declared bin resolves and ships
  285. ├── package.json
  286. └── README.md
  287. ```
  288. ## Latest Updates
  289. ### v2.2.1 (2026-10-21)
  290. - **Fixed the documented `npx claude-skill-antivirus` command**, which failed with "could not determine executable to run" because no bin matched the package name. `claude-skill-antivirus` is now a bin alias alongside `skill-install` and `claude-skill-av`, and a test asserts every declared bin resolves, runs `--version`, and ships in the tarball.
  291. - `--help` now reports the alias you actually typed instead of always printing `skill-install`.
  292. - npm publishing moved to [trusted publishing](https://docs.npmjs.com/trusted-publishers) (OIDC), so releases are signed with provenance and no long-lived npm token exists in CI.
  293. ### v2.2.0 (2026-10-21)
  294. - **Security fix ([#33](https://github.com/claude-world/claude-skill-antivirus/issues/33))**: the scanner now walks the **whole skill directory** instead of reading only `SKILL.md`, so hidden code in `scripts/`, nested folders, or a poisoned `__pycache__/*.pyc` can no longer score a false SAFE 100/100. Reported by [@nedlir](https://github.com/nedlir); directory-walk foundation by [@Yahhi](https://github.com/Yahhi) ([#34](https://github.com/claude-world/claude-skill-antivirus/pull/34)).
  295. - New **File Artifact Scanner** (10th engine) flags compiled/binary/archive files by magic bytes, and a PEP 552 unchecked-hash `.pyc` is CRITICAL; skipped dirs, unfollowed symlinks, truncated files, and scan-limit hits are now reported as findings.
  296. - GitHub repository, `tree/` and `blob/.../SKILL.md` URLs walk the whole skill via the GitHub API; `.zip` inputs are hardened (zip-slip, encrypted/oversized entries, size caps).
  297. - Installer hardening: refuses zip-slip paths and incompletely scanned skills, and writes binaries byte-for-byte.
  298. - Fixed ReDoS across the engines (a crafted long line could hang the scanner) and a crash on SKILL.md files containing a `| name |` table.
  299. - Pattern precision tuned for real code now that scripts are scanned. Skills that ship scripts will show more findings than in 2.1.x; review them.
  300. - Verified compatibility with Claude Code Opus 5.5, Sonnet 5.5, Haiku 5.5, and Fable 5.1.
  301. - Node.js >= 22; upgraded dependencies (commander 15, inquirer 14, chalk 6, ora 9, yaml 2.9, adm-zip 0.6); removed `node-fetch` in favor of built-in `fetch`.
  302. ### v2.1.0 (2026-03-13)
  303. - Verified compatibility with Claude Code Opus 4.6
  304. - Updated documentation and metadata
  305. ### v2.0.1
  306. - Separated capability warnings from actual threats in permission scanner
  307. - Fixed array format handling in `allowed-tools`
  308. ### v2.0.0
  309. - Added 4 new scanning engines: MCP Security, SSRF, Dependency, and Sub-agent scanners (total: 9 engines)
  310. - Added i18n support (English + Traditional Chinese)
  311. - Added batch scanner for SkillsMP platform
  312. - Scanned all 71,577 skills on SkillsMP
  313. ### v1.0.0
  314. - Initial release with 5 core scanning engines
  315. - CLI installer with interactive prompts
  316. ## Related Projects
  317. - [cf-browser](https://github.com/claude-world/cf-browser) - Open-source Cloudflare Browser Rendering proxy with 9 MCP tools for Claude Code
  318. - [claude-world.com](https://claude-world.com) - Claude Code advanced usage community
  319. ## Contributing
  320. Contributions are welcome! Please feel free to submit issues and pull requests.
  321. ### Adding New Detection Patterns
  322. Each scanner is modular. To add new patterns:
  323. 1. Find the appropriate scanner in `src/scanner/`
  324. 2. Add your pattern to the relevant array
  325. 3. Include: `pattern`, `risk`, `title`, `description`
  326. ## License
  327. MIT
  328. ## Author
  329. Lucas Wang <support@claude-world.com>
  330. ## Links
  331. - [GitHub Repository](https://github.com/claude-world/claude-skill-antivirus)
  332. - [npm Package](https://www.npmjs.com/package/claude-skill-antivirus)
  333. - [Report Issues](https://github.com/claude-world/claude-skill-antivirus/issues)

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.