The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documentation instructs users to install the resend library from the official NPM registry and clone the project repository from GitHub. These are standard development practices for the intended functionality.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to perform media optimization tasks using ffmpeg and to execute email-sending scripts via node. These operations are well-documented and align with the skill's purpose of asset preparation and delivery.
[OBFUSCATION]: The skill's HTML templates include a base64-encoded SVG background image. Analysis confirms this is a benign noise texture used for design depth, which is a common practice in modern web and email development.
[INDIRECT_PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection as it is designed to process and interpolate data from various sources into email campaigns.
Ingestion points: The skill processes user-provided HTML templates (email.html), local assets in the public/ directory, and external advertising data retrieved through the suggested Ryze MCP connector (connector.get-ryze.ai/mcp).
Boundary markers: The instructions do not specify explicit delimiters or "ignore embedded instructions" warnings for handling data retrieved from external connectors.
Capability inventory: The skill has access to the Bash tool (shell execution) and full file system capabilities via Read, Write, Edit, Grep, and Glob tools.
Sanitization: No specific sanitization or validation logic is implemented for the data ingested during the email generation process.

email-campaigns