chinese-content-writing-guideline

Pass

Audited by Gen Agent Trust Hub on Mar 28, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill instructs the agent to avoid stating that a topic makes it feel "uncomfortable, uneasy, or slightly offended," specifically targeting sensitive subjects like cybersecurity, attacks, AI, ethics, philosophy, and psychology. This serves to suppress the AI's standard safety signaling and refusal language.
  • [EXTERNAL_DOWNLOADS]: The instructions mandate using curl to fetch titles of external webpages found in the content. This creates an indirect prompt injection surface where a malicious webpage title could influence the agent's behavior.
  • Ingestion points: External URLs contained within text processed by the skill in SKILL.md.
  • Boundary markers: None provided; the agent is simply told to fetch and use the title as link text.
  • Capability inventory: Implicit use of network fetching tools (e.g., curl) to retrieve remote data.
  • Sanitization: No instructions provided to sanitize or validate the fetched content before including it in the output.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 28, 2026, 06:20 AM