The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes user-supplied tokens and interpolates them into markdown templates which are subsequently written to the filesystem. This creates an indirect prompt injection surface.
Ingestion point: The tokens argument in main.cjs and SKILL.md.
Boundary markers: No specific delimiters or "ignore embedded instructions" markers are applied to the injected token values within the templates.
Capability inventory: The skill utilizes mcp__filesystem__write_file and Write to save the rendered output to the local project structure.
Sanitization: A sanitizeTokenValue function is implemented in main.cjs and SKILL.md to remove HTML tags and template delimiters ({{, ${) from the input tokens.

template-renderer