create-viral-content
Pass
Audited by Gen Agent Trust Hub on Mar 9, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The
SKILL.mdfile contains an 'AUTOMATIC ACTIVATION' section with directive language intended to override the agent's default behavior whenever specific keywords or contexts are encountered. Patterns like 'Use this skill whenever...', 'Apply to any AI-generated content...', and 'triggers on...' are used to force the agent into a specific operational mode. - [PROMPT_INJECTION]: The skill includes instructions to 'Kill these on sight' and 'Zero tolerance for blacklisted phrases' regarding 'AI Tells'. These are meta-instructions that command the model to ignore its training and natural output characteristics, which is a form of behavioral override.
- [SAFE]: The skill references several well-known research sources and platforms such as BuzzSumo, Netflix, and Outbrain for statistical basis. These are documented as external research and do not involve runtime network requests or untrusted code execution.
- [SAFE]: The
humanize-integration.mdfile describes integration with an external skill (humanize-writing). This reference is treated as an informative instruction about the agent environment's capabilities rather than a security risk.
Audit Metadata