The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill uses scripts/export-pdf.sh to dynamically generate a temporary Node.js script (export-slides.mjs) which is then executed using node. This script automates a headless browser to capture slide screenshots for PDF assembly.
[EXTERNAL_DOWNLOADS]: The skill triggers the installation of third-party dependencies during its workflow, including the python-pptx and Pillow libraries via pip, and the playwright and vercel packages via npm/npx. It also initiates the download of the Chromium browser binary for Playwright operations. These are well-known tools used for the skill's primary features.
[COMMAND_EXECUTION]: The skill frequently executes shell scripts and system commands to perform its duties. This includes running bash scripts/deploy.sh to host files on Vercel, bash scripts/export-pdf.sh for PDF creation, and python scripts/extract-pptx.py for PowerPoint content extraction. It also uses the open command to launch the generated presentations in the user's browser.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes external, untrusted content from PowerPoint (.pptx) files and user-provided image directories to generate presentation code.
Ingestion points: The scripts/extract-pptx.py script reads text and metadata from user-supplied PowerPoint files. The skill also scans user-provided folders for image assets.
Boundary markers: None explicitly implemented in the generation prompt; however, the skill is instructed to present extracted content to the user for confirmation before proceeding to the generation phase.
Capability inventory: The skill has the ability to write to the local filesystem, execute shell scripts, and perform network operations via the Vercel deployment tool.
Sanitization: Relies on user review of extracted summaries and titles to identify potential malicious content before it is processed by the agent into the final HTML output.

frontend-slides