data-research

Pass

Audited by Gen Agent Trust Hub on Aug 29, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill represents a significant attack surface for indirect prompt injection because it ingests and processes untrusted data from multiple external channels.
  • Ingestion points: Phase 2 identifies ingestion from email bodies, web search results (public filings, press releases), API responses, and file attachments (PDF, HTML).
  • Boundary markers: The skill instructions do not specify the use of delimiters or 'ignore instructions' warnings when processing these external sources.
  • Capability inventory: The skill possesses extensive capabilities including writing to pages (put_page), writing raw files (put_raw_data), uploading files (file_upload), and performing network searches (search).
  • Sanitization: While the 'EXTRACTION INTEGRITY RULE' in Phase 4 addresses data accuracy and hallucination prevention, it does not provide mechanisms for sanitizing adversarial instructions contained within the ingested sources.
  • [EXTERNAL_DOWNLOADS]: The skill is designed to fetch content from various remote network locations.
  • Evidence: Phase 2 explicitly mentions performing web searches, querying APIs, and downloading attachments from emails.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 29, 2026, 06:18 PM
Security Audit — agent-trust-hub — data-research