nuxt-content
Fail
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: HIGHPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [Indirect Prompt Injection] (HIGH): The skill creates a high-severity attack surface by requiring the agent to ingest untrusted external data and use it for code implementation.
- Ingestion points: The URL https://content.nuxt.com/llms.txt referenced in SKILL.md.
- Boundary markers: Absent. The skill provides a prompt template that lacks instructions to ignore or treat embedded directives in the documentation as data rather than instructions.
- Capability inventory: The skill is used for 'implementing features' and 'writing code', giving it significant side-effect potential.
- Sanitization: Absent. The content is passed directly to the LLM context.
- [Prompt Injection] (MEDIUM): The instructions utilize absolute and aggressive language ('ABSOLUTELY FORBIDDEN', 'MANDATORY', 'CRITICAL') to override the agent's internal reasoning and default safety prioritization.
- [External Downloads] (LOW): The skill systematically fetches data from a non-whitelisted external domain (content.nuxt.com). While fetching text, this creates a dependency on an external source for agent behavior.
Recommendations
- AI detected serious security threats
Audit Metadata