web-fetch
Warn
Audited by Snyk on Feb 20, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 1.00). The SKILL.md explicitly fetches arbitrary http/https URLs via curl and a Bun fetch script (e.g., "curl -s """ and "bun ~/.claude/skills/web-fetch/fetch.ts """) to convert third‑party HTML into Markdown for the agent to read, so untrusted public web content can directly influence the agent's subsequent interpretation and actions.
MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).
- Potentially malicious external URL detected (high risk: 1.00). The skill fetches arbitrary http/https URLs at runtime (e.g., the command pattern curl -s "" | html2markdown ...) and injects that remote page content into the agent context, so fetched content can directly control prompts.
Audit Metadata