The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use a shell command to open files on the host system.
Evidence: The documentation in SKILL.md explicitly provides the command Bash(open path/to/show-me-{description}.html) to display visual results.
[DYNAMIC_EXECUTION]: The skill generates local files at runtime and then executes a command to view them, which is a form of dynamic content execution.
Evidence: In SKILL.md, the agent is instructed to "write one focused HTML file" and then "open it for the user". This allows the agent to create and then trigger the execution of arbitrary HTML and potentially JavaScript content.
[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted conversation data to generate its visualizations, creating a surface for injection.
Ingestion points: The skill takes the "current topic of conversation" and user-provided diagrams or code as input (SKILL.md).
Boundary markers: No specific delimiters or instructions to ignore embedded commands are present in the skill's instructions.
Capability inventory: The agent possesses the capability to write files to the local disk and execute shell commands (SKILL.md).
Sanitization: The instructions lack guidance on sanitizing or escaping conversation data before including it in the generated HTML or diagrams.

show-me