The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from user-provided notes and PowerPoint files to generate HTML and JavaScript presentations. This creates a potential vector where malicious scripts or instructions hidden in the source files could be executed in the generated output. -- Ingestion points: Content extraction from .pptx files in Phase 4 and user-provided notes or topics collected in Phase 1. -- Boundary markers: The provided logic does not include explicit delimiters or instructions to the agent to treat the extracted content as untrusted data during the generation phase. -- Capability inventory: The skill performs local file system writes and executes Python scripts to process image and presentation data. -- Sanitization: The instructions lack a requirement for the agent to escape or sanitize the extracted text strings before they are interpolated into the HTML presentation templates.

frontend-slides