The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt or template injection because it performs raw string substitution of variables into MJML templates.
Ingestion points: The variables dictionary in the MjmlTemplateRenderer.RenderTemplateAsync method and the UserEmailComposer class (SKILL.md).
Boundary markers: Absent. The skill uses a simple regex replacement pattern {{variable}} to merge data into the markup.
Capability: The skill generates HTML email bodies intended to be sent to recipients and rendered in email clients.
Sanitization: Absent. The SubstituteVariables implementation does not perform HTML encoding or sanitization on the input values, meaning malicious MJML or HTML tags provided in variables (like a user's name or organization name) will be processed by the rendering engine.
[DYNAMIC_EXECUTION]: The skill assembles and executes MJML code dynamically at runtime based on template strings and variable values.
Evidence: The RenderTemplateAsync method constructs a complete MJML string by concatenating layout and content templates and then calls _mjmlRenderer.RenderAsync(processedMjml, ...) to compile the resulting string into HTML.
[EXTERNAL_DOWNLOADS]: The skill references and utilizes an external package for its core functionality.
Evidence: The installation instructions include a command to add the Mjml.Net package from the NuGet registry.

mjml-email-templates