medical-research-literature-reader-pro

Pass

Audited by Gen Agent Trust Hub on May 10, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is primarily instructional and provides a rigorous framework for critical appraisal of scientific literature. It does not contain any executable scripts, binary files, or network-enabled commands.
  • [PROMPT_INJECTION]: The skill contains defensive instructions to prevent biased analysis. Specifically, the 'Behavioral Rules' in SKILL.md instruct the agent to ignore user requests for 'biased analysis' (e.g., strength-only reviews) and instead provide a balanced critique. This is a positive security and integrity feature.
  • [DATA_INGESTION]: As an indirect prompt injection surface, the skill is designed to ingest and process untrusted external data (medical papers, PDFs, and abstracts).
  • Ingestion points: SKILL.md (Input Handling section) defines inputs including full PDFs and PMIDs.
  • Boundary markers: The 'Mandatory Output Template' (SKILL.md) and 'Interpretation Safety Rules' (references/reporting_style.md) provide structural boundaries for the AI's response.
  • Capability inventory: No hazardous capabilities such as subprocess execution, filesystem writes, or network exfiltration were found in the provided files.
  • Sanitization: The skill enforces strict rules against fabrication and overclaiming (SKILL.md, Behavioral Rules), ensuring the agent remains grounded in the provided text.
Audit Metadata
Risk Level
SAFE
Analyzed
May 10, 2026, 12:19 AM