contradiction-finder

Pass

Audited by Gen Agent Trust Hub on Sep 10, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches research paper metadata and abstracts from established academic services including CrossRef (api.crossref.org), Semantic Scholar, and arXiv to populate its analysis set.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data in the form of paper abstracts which could theoretically contain instructions intended to influence the agent.
  • Ingestion points: External content is ingested through API calls to CrossRef and Semantic Scholar in Step 1 of the Protocol.
  • Boundary markers: The instructions mandate the use of source paper attribution and verbatim excerpts in Step 2 to maintain a clear distinction between source text and agent analysis.
  • Capability inventory: The skill's capabilities are restricted to text-based extraction, semantic comparison, and output formatting; it does not utilize file-system writes, network exfiltration, or shell execution tools.
  • Sanitization: The agent is explicitly instructed to verify that DOIs resolve to real papers via CrossRef before processing, preventing the use of fabricated content.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 10, 2026, 10:15 PM
Security Audit — agent-trust-hub — contradiction-finder