⚡ Frisk

Security verification for ClawHub skills. Scan any skill — by local path or ClawHub slug — against 9 continuously-updated threat intelligence sources using 7 autonomous security checks.

What it does

Frisk scans a skill directory for security issues before you install it. It checks dependencies against known vulnerability databases, scans for hardcoded secrets, detects indicators of compromise, and looks for prompt injection vectors — all without sending your data anywhere.

One command to scan any ClawHub skill:

frisk scan weather-forecast

That downloads the skill from ClawHub, scans it, shows results, and cleans up. No manual steps.

Or scan a local skill directory:

frisk scan ./my-skill

When to use this skill

Use Frisk when you are about to install a skill from ClawHub and want to verify it is safe. Also use it when developing your own skills — run a scan before publishing to catch issues early.

Parameters

• target (required) — A local directory path or a ClawHub skill slug (e.g. weather-forecast). If a slug is provided, the skill is downloaded temporarily, scanned, and removed.
• checks (optional) — Comma-separated list of checks to run: dep-scan, static-analysis, secret-scan, yara-scan, ioc-match, behavioral, prompt-inject. Default: all 7.
• json (optional) — Output results as JSON for programmatic use.

Return value

Frisk outputs a structured report with:

• verdict — pass, warn, or fail
• findings — Array of issues found, each with severity (critical, high, medium), description, and file location
• report_id — Short ID for later retrieval via frisk report \x3Cid>

Exit codes: 0 = pass, 1 = warn, 2 = fail

Checks

Threat intel sources (9, continuously synced)

CISA KEV · OSV (npm + PyPI) · EPSS · MalwareBazaar · URLhaus · ThreatFox · Feodo Tracker · YARA Rules · Semgrep Rules

Run frisk sync to refresh the intel cache.

Security & Privacy

• No data leaves your machine. All scanning happens locally. No telemetry, no phone-home, no analytics.
• Downloaded skills are sandboxed. When scanning by slug, the skill is downloaded to a 0700-permission temp directory, all files have execute permissions stripped before scanning, and npm postinstall scripts are suppressed. The skill is deleted after scanning.
• Credentials stay local. Frisk reads environment variables for configuration but never transmits them.

External endpoints

Frisk downloads threat intel feeds from these public sources during frisk sync:

During frisk scan, no network requests are made. All intel is local.

Local files read

• ~/.frisk/intel/ — Threat intel cache
• Skill directory passed as scan target

Local files written

• ~/.frisk/intel/ — Synced threat intel data
• ~/.frisk/reports/ — Scan reports (JSON)
• ~/.frisk/venv/ — Python virtual environment for scan checks
• ~/.frisk/frisk.log — Scan log output

Trust statement

By using Frisk, you trust the threat intel sources listed above to provide accurate vulnerability and IOC data. No skill code or scan targets are transmitted to any external service. Install it only if you trust the Low Watt Labs project and the listed intel sources.

Install

npm install -g @lowwattlabs/frisk

First run automatically creates a Python venv and syncs threat intel. After that, frisk scan works with zero configuration.

License

MIT-0 — same as all ClawHub skills.