The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill manages the installation of slide production toolchains using well-known package managers. It provides commands for brew install marp-cli, brew install pandoc, brew install tectonic, npm i -g @marp-team/marp-cli, and pip3 install python-pptx. Additionally, the tectonic LaTeX engine is noted to download necessary packages from the network during its initial execution. These downloads are associated with official repositories and well-known developer tools.\n- [COMMAND_EXECUTION]: The renderer_agent is designed to execute shell commands to build, verify, and open slide deck files. These include marp, pandoc, tectonic, and macOS system utilities like open (to display artifacts) and mdls (to verify page counts). The use of these commands is consistent with the skill's utility and is clearly documented for the user.\n- [REMOTE_CODE_EXECUTION]: The figure_maker_agent generates Python, Mermaid, and TikZ scripts which are executed at runtime to render figures. This dynamic execution (e.g., python3 figures/src/W<N>_fig<k>.py) is a core feature for creating reproducible, theme-aligned graphics from data and descriptions provided by the professor.\n- [PROMPT_INJECTION]: The skill processes untrusted input in the form of lecture notes and slide outlines, presenting a surface for indirect prompt injection.\n
Ingestion points: Lecture notes and slide outlines provided by the user (or sourced from a lesson-builder skill) serve as the primary input.\n
Boundary markers: The skill employs structured markdown templates and HTML comment directives (e.g., Marp directives) to maintain separation between instructions and content.\n
Capability inventory: The skill possesses file system write access and the ability to execute shell commands and Python scripts for rendering purposes.\n
Sanitization: Content is mapped into predefined templates like templates/deck_source_template.md, which provides structural isolation for the processed text.

deck-studio