light-spotlight-render
Pass
Audited by Gen Agent Trust Hub on Nov 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input to generate HTML content, creating a surface for injection attacks.
- Ingestion points: User-provided parameters such as
label_text,mask_color,text_color, andbackground_colorare collected inSKILL.mdand passed toscripts/render_light_spotlight.py. - Boundary markers: None. The skill does not explicitly warn the agent to ignore instructions embedded in the
label_textor validate the format of the color strings. - Capability inventory: The skill uses
python3to execute a local script which performs file system writes (output_path.write_text) to create an HTML file. - Sanitization: While
label_textis sanitized viahtml.escape()and slugified for the filename, the color parameters (mask_color,text_color,background_color) are only stripped of whitespace. An attacker could provide values likered; } body { background: url(https://evil.com/leak); }to inject arbitrary CSS or attempt to break out of the SVG/style context if the template is not strictly guarded.
Audit Metadata