The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided data from a JSON file to populate the Kanban board.
Ingestion points: The kanban.json source file is read and parsed by the renderer script.
Boundary markers: Instructions explicitly direct the agent to treat the file as a strict source of truth and avoid inventing content.
Capability inventory: The skill is capable of writing HTML files to the disk and launching the default system browser.
Sanitization: The JavaScript template uses textContent-based escaping to prevent XSS, and the data injection into the HTML body is protected against script tag breakout using character substitution.
[COMMAND_EXECUTION]: The tool uses execFileSync to call platform-specific file openers like open or xdg-open. This is the intended behavior for viewing the generated report and follows best practices by avoiding shell interpolation.

kanban