skills/danangjoyoo/nerd/nerd-patrol/Gen Agent Trust Hub

nerd-patrol

Pass

Audited by Gen Agent Trust Hub on Aug 17, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external content (code and configurations), which inherently introduces an attack surface for indirect instructions.
  • Ingestion points: The skill body in SKILL.md specifies the examination of "confirmed code, pull request, module, application, endpoint, dependency, or configuration scope."
  • Boundary markers: The skill relies on structured templates (Patrol Scope, Security Finding) but does not define explicit delimiters or instructions to ignore embedded prompts in audited data.
  • Capability inventory: Instructions mention an "execute endpoint" and permit the agent to "change code," suggesting the presence of file system write or shell execution capabilities.
  • Sanitization: The skill implements safety measures by requiring the use of "harmless" payloads and "sanitized" requests, which serves to mitigate the risk of unintended code execution or exfiltration.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 17, 2026, 01:24 AM
Security Audit — agent-trust-hub — nerd-patrol