markstream-nuxt
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a workflow for rendering potentially untrusted content (AI chat streams, Markdown documents) through the
MarkdownRendercomponent. While this is a standard feature, it represents an injection surface if improperly configured. - Ingestion points: The component processes
contentandnodesvariables for UI rendering inSKILL.md. - Boundary markers: The instructions explicitly mandate
html-policy="safe"and Mermaid strict mode as the default configuration to prevent malicious content execution. - Capability inventory: The skill does not define dangerous capabilities such as file system writes, network exfiltration, or arbitrary command execution for the agent.
- Sanitization: Security is maintained by the underlying library's sanitization logic, which the skill configures to a safe state by default.
Audit Metadata