customer-signal-generator

Fail

Audited by Gen Agent Trust Hub on Oct 27, 2026

Risk Level: CRITICALEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Automated security scans identified a malicious URL (phishing) within the provided sample documentation. The URL "https://ssojet.com/security-vendors/snyk-alternatives/" is contained in the "references/examples/snyk-sample-report.html" file, which is included as a sample output.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from public web pages and job postings, creating a surface for indirect prompt injection. \n
  • Ingestion points: Web research results (Step 1) and Clay company data (Step 2). \n
  • Capability inventory: The skill writes HTML/YAML files and executes a local Python script to render reports. \n
  • Sanitization: The "scripts/render_report.py" script uses "html.escape()" to provide basic XSS protection, but the agent's reasoning process remains susceptible to manipulation by instructions embedded in external content.
  • [COMMAND_EXECUTION]: The skill executes a local Python script to process data and generate HTML reports. Evidence: "SKILL.md" contains instructions to execute "python3 scripts/render_report.py" with specific content and signal file arguments.
Recommendations
  • Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Oct 27, 2026, 07:44 PM