The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a shell command to open generated HTML files. It uses a template format that includes a dynamic placeholder for the filename.
Evidence: The instruction Bash(open path/to/show-me-{description}.html) directs the agent to invoke the system's open utility with a path derived from user-provided descriptions.
[INDIRECT_PROMPT_INJECTION]: The skill processes the 'current topic of conversation' to generate visualizations. Since this data can originate from untrusted sources (e.g., website content or file data being discussed), it presents an injection surface.
Ingestion points: Conversation context used to generate diagrams (SKILL.md).
Boundary markers: None provided to distinguish between visualization instructions and potentially malicious data within the conversation.
Capability inventory: Capability to write files and execute Bash(open ...) (SKILL.md).
Sanitization: No explicit sanitization or validation of the {description} placeholder or the HTML content is defined in the instructions.
[DYNAMIC_EXECUTION]: The skill involves the runtime creation of HTML files followed by an instruction to open them, which constitutes dynamic execution of generated content.
Evidence: The skill guides the agent to 'write one focused HTML file' and then 'open it for the user'.

show-me