The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its rendering script.\n
Ingestion points: The --mask-color, --text-color, and --background-color arguments in scripts/render_light_spotlight.py which can be populated by untrusted external data.\n
Boundary markers: No delimiters or sanitization are applied to these color parameters before they are interpolated into the HTML template context.\n
Capability inventory: The script has the capability to write to the local filesystem (output_path.write_text), enabling the deposition of malicious payloads.\n
Sanitization: While label_text is escaped using html.escape, the color inputs are inserted raw, allowing a malicious source to inject HTML or JavaScript (XSS) into the generated file.\n- [COMMAND_EXECUTION]: The Python script allows specifying an arbitrary file path for the output animation.\n
The --output argument is used directly in Path.resolve() without restricting the write operation to a specific safe directory, potentially allowing the agent to overwrite system or sensitive files if prompted to do so.\n- [EXTERNAL_DOWNLOADS]: The generated HTML template includes an external resource.\n
Fetches the GSAP library from Cloudflare's public CDN to handle animations.

light-spotlight-render