The Agent Skills Directory

[SAFE]: The skill documentation and example code demonstrate standard React component integration practices without any signs of malicious activity.
[INDIRECT_PROMPT_INJECTION]: The skill involves rendering untrusted markdown content from an LLM, which presents a potential injection surface. However, the instructions provide adequate security measures.
Ingestion points: The content property of the MarkdownRender component in SKILL.md.
Boundary markers: The documentation explicitly instructs users to apply security settings when handling content.
Capability inventory: The skill is limited to frontend UI rendering and does not invoke system commands, file system writes, or network tools.
Sanitization: The workflow mandates the use of htmlPolicy="safe" and strict Mermaid diagram rendering to prevent cross-site scripting (XSS) or other content-based attacks.
[EXTERNAL_DOWNLOADS]: The skill references a legitimate frontend package markstream-react which is consistent with the author's own software repository and poses no security threat.

markstream-react