The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill navigates to external websites provided by the user and uses browser_snapshot to analyze their content for generating marketing copy. This creates a surface where a malicious website could embed instructions designed to influence the agent's output.
Ingestion points: Website content retrieved via browser_navigate and browser_snapshot in SKILL.md.
Boundary markers: None are specified; the agent is instructed to read the site content directly to write descriptions.
Capability inventory: The skill uses browser_take_screenshot, browser_evaluate, and bash to create directories and write marketing files.
Sanitization: No sanitization or filtering of the retrieved web content is mentioned before it is processed by the agent.
[COMMAND_EXECUTION]: The skill executes shell commands (mkdir -p) to set up an output directory within the project root. While this is a standard operational task, it involves interpolating user-provided or environment-derived paths into a shell environment.
[CREDENTIALS_UNSAFE]: The instructions explicitly guide the agent to ask the user for session cookies or login credentials if a page is behind authentication. Although the skill includes a warning not to store or hardcode these credentials, the act of processing active session tokens or passwords through the model's context represents a potential data exposure risk in logs or during processing.

generate-marketing-screenshots