trump-perspective
Fail
Audited by Gen Agent Trust Hub on Apr 10, 2026
Risk Level: CRITICALEXTERNAL_DOWNLOADSPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The research documentation in "references/research/03-expression-dna.md" includes a link to "https://ijels.com/upload_document/issue_files/70IJELS-1042024-TheArt.pdf", which has been explicitly identified by security scanners as a malicious URL associated with botnet activity.
- [PROMPT_INJECTION]: The skill utilizes directive language in "SKILL.md" designed to force specific agent behavior and priority execution ("激活确认 (优先执行)"). These patterns are used to mandate a persona and response style, which can be leveraged to attempt overriding base safety and behavioral constraints.
- [DATA_EXFILTRATION]: The skill implements an "Agentic Protocol" that requires the agent to perform web searches and process external findings. This design establishes a significant surface for indirect prompt injection, as the agent is instructed to ingest, summarize, and act upon potentially untrusted external content without defined sanitization or boundary markers.
- [SAFE]: The installation procedure using "npx skills add" and references to official developer platforms and established version control repositories are standard and do not present an immediate security concern.
Recommendations
- AI detected serious security threats
- Contains 2 malicious URL(s) - DO NOT USE
Audit Metadata