markstream-custom-components
Pass
Audited by Gen Agent Trust Hub on Aug 11, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The content is primarily instructional, guiding the agent on how to implement custom component renderers for Markdown nodes without introducing malicious scripts or hidden behaviors.
- [PROMPT_INJECTION]: The skill handles untrusted Markdown input, which constitutes an indirect prompt injection surface (Category 8). The skill mitigates this risk by providing specific security instructions. Ingestion points: The
markdownproperty passed to theMarkdownRendercomponent inSKILL.md. Boundary markers: Explicit use of thehtmlPolicy="safe"attribute in rendering examples. Capability inventory: Frontend component rendering and dynamic component registration within React, Vue, Svelte, and Angular frameworks. Sanitization: Detailed notes advising the agent to "Keep safe HTML enabled" and avoid passing unsanitized attributes into host components.
Audit Metadata