The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes external PowerPoint files which may contain untrusted content, creating a surface for indirect prompt injection.
Ingestion points: The extract_pptx function in SKILL.md reads text, images, and notes from user-provided PowerPoint files.
Boundary markers: No explicit delimiters or instructions are provided to the agent to disregard instructions that might be embedded within the extracted slide content during confirmation or generation phases.
Capability inventory: The skill has the ability to write HTML files to the disk and use the open command to launch them in a browser, which could be exploited if the agent is influenced by malicious instructions in a processed document.
Sanitization: No sanitization or validation of the content extracted from the PowerPoint files is performed before processing.
[COMMAND_EXECUTION]: The skill requires the agent to execute a Python script locally to parse documents and uses the open command to display the generated presentation in the user's browser.
[EXTERNAL_DOWNLOADS]: Generated presentations are configured to fetch font assets and stylesheets from api.fontshare.com, which is a well-known third-party typography service.

frontend-slides