discover-phenomenon-and-gap

Pass

Audited by Gen Agent Trust Hub on Jul 12, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to ingest and analyze external, untrusted content from scientific papers.
  • Ingestion points: The skill reads abstracts and full-text content from external sources via tools like literature-search-openalex, literature-search-arxiv, literature-search-biorxiv, and pubmed-database (identified in SKILL.md).
  • Boundary markers: Absent. There are no explicit instructions or delimiters defined to prevent the agent from interpreting instructions embedded within the research papers as its own commands.
  • Capability inventory: The skill has the ability to write to local files (temp_limitations_log.md) and execute subsequent search queries based on extracted content.
  • Sanitization: Absent. The workflow instructs the agent to extract "Future Works" and "Usage Notes" directly from external text without validation or sanitization.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 12, 2026, 11:44 AM
Security Audit — agent-trust-hub — discover-phenomenon-and-gap