hypothesis-generation

Pass

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: LOWPROMPT_INJECTIONNO_CODE
Full Analysis
  • [Indirect Prompt Injection] (LOW): The skill ingests untrusted external literature through WebFetch and WebSearch tools, creating an attack surface for indirect prompt injection.
  • Ingestion points: Step 2 in SKILL.md (Conduct Comprehensive Literature Search) utilizes external web and PubMed data.
  • Boundary markers: No specific boundary markers or delimiters are used to isolate external content from instructions.
  • Capability inventory: The skill is limited to reasoning and generating text based on templates; it does not have the capability to execute commands, write to the filesystem, or perform network requests beyond the initial search.
  • Sanitization: No sanitization or filtering of external content is specified.
  • [Data Exposure & Exfiltration] (SAFE): No sensitive file access, hardcoded credentials, or unauthorized network activity detected. Tool use is limited to literature search capabilities.
  • [Remote Code Execution] (SAFE): No dynamic code execution, library injection, or remote script downloads identified.
Audit Metadata
Risk Level
LOW
Analyzed
Feb 16, 2026, 09:52 AM