The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use the open command to launch the generated HTML file in a web browser for previewing. Executing local files generated from untrusted web data is a potential attack vector.
[EXTERNAL_DOWNLOADS]: The generated report fetches the Chart.js library from the JSDelivr CDN. This is a reference to a well-known service used for data visualization.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface due to its data-handling flow:
Ingestion points: Untrusted competitor data, market share figures, and product parameters are retrieved via the WebSearch tool as described in Step 2 of SKILL.md.
Boundary markers: Absent. The instructions do not define delimiters or warnings to prevent search results from influencing the report's structure or the agent's behavior during generation.
Capability inventory: The skill possesses the ability to write files to the local workspace and execute shell commands (open).
Sanitization: Absent. There is no instruction to validate, escape, or filter the content retrieved from web searches before interpolating it into the HTML and JavaScript templates.
[DYNAMIC_EXECUTION]: The skill dynamically generates an executable HTML/JavaScript file at runtime. This file incorporates content retrieved from the web (external search results). If a searched website contains malicious scripts or cross-site scripting (XSS) payloads, they would be written into the local report file and executed when the user or agent opens the file in a browser.

competitive-landscape