The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of the markstream-svelte package, which is currently in a beta state. The instructions correctly prompt the user for explicit approval before proceeding with the installation of this external dependency.
[PROMPT_INJECTION]: The skill is designed to process and render untrusted markdown data, creating a surface for indirect prompt injection.
Ingestion points: The content property in the MarkdownRender component example (SKILL.md).
Boundary markers: The skill recommends the use of htmlPolicy="safe" and strict defaults for Mermaid diagrams to isolate rendered content.
Capability inventory: The skill is focused on UI rendering; no subprocess execution, network exfiltration, or filesystem writes were identified in the instructions (SKILL.md).
Sanitization: The skill explicitly advises using a safe HTML policy and strict security defaults for integrated components to filter malicious inputs.

markstream-svelte