The Agent Skills Directory

[PROMPT_INJECTION]: The skill implements an 'Extract' workflow in references/extractors/from-website.md, references/extractors/from-pdf.md, and references/extractors/from-video.md that ingests content from untrusted external sources.
Ingestion points: Website URLs, PDF brand guides, and video keyframes are processed to synthesize a style_prompt_full field.
Boundary markers: The skill lacks explicit boundary markers or instructions to ignore embedded commands within the source data.
Capability inventory: The generated output is directly interpolated into prompts for the HeyGen Video Agent (references/connectors/heygen-video-agent.md), which executes instructions to create AI video content.
Sanitization: No sanitization or validation of the extracted natural language content is performed before interpolation.
[COMMAND_EXECUTION]: The references/connectors/html-slides.md connector generates standalone HTML slide presentations with embedded CSS and JavaScript based on design system metadata. If the design system is extracted from a malicious source, it could lead to Cross-Site Scripting (XSS) or the execution of unauthorized client-side scripts when the slides are opened in a browser.
[EXTERNAL_DOWNLOADS]: The documentation in references/connectors/html-slides.md references an external, unverified GitHub repository (github.com/zarazhangrui/frontend-slides) as a recommended framework. While this is a documentation reference and not an automated dependency installation, it directs users to third-party code without integrity verification.

visual-style