cell-host-and-microbe

Pass

Audited by Gen Agent Trust Hub on Sep 24, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill is composed entirely of natural language instructions and lacks any executable code, shell commands, or network operations. It correctly advises users to follow official guidelines and manage data/code through standard repositories.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided manuscript data, which is an untrusted input source. However, the risk is mitigated by the complete absence of dangerous capabilities.
  • Ingestion points: User manuscripts analyzed for journal fit.
  • Boundary markers: None.
  • Capability inventory: No file-write, network, or command execution tools are utilized.
  • Sanitization: None.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 24, 2026, 11:33 AM
Security Audit — agent-trust-hub — cell-host-and-microbe