icp-onboarding

Fail

Audited by Gen Agent Trust Hub on Apr 30, 2026

Risk Level: HIGHCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute a shell command npx tsx scripts/scrape-website.ts --url=[URL] using input directly from the user. This creates a high risk of command injection, where a malicious user could provide a URL containing shell metacharacters (e.g., ;, &&, or |) to run arbitrary system commands.\n- [EXTERNAL_DOWNLOADS]: The scripts/scrape-website.ts script uses the fetch API to download content from arbitrary external websites. This could be abused for Server-Side Request Forgery (SSRF) to scan internal networks or access sensitive local metadata services.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection by scraping external website content and presenting it to the agent for processing.\n
  • Ingestion points: Web content fetched from remote URLs is saved to /tmp/scrape.json.\n
  • Boundary markers: There are no instructions or delimiters to isolate the scraped content from the agent's primary instructions.\n
  • Capability inventory: The skill allows shell command execution and file writes.\n
  • Sanitization: The script only performs basic HTML tag removal, leaving any malicious natural language instructions intact for the agent to process.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Apr 30, 2026, 10:21 AM
Security Audit — agent-trust-hub — icp-onboarding