The Agent Skills Directory

[COMMAND_EXECUTION]: The skill employs the Bash tool to execute standard local utilities including xelatex, bibtex, pdftoppm, pdfinfo, and pdf2svg. These tools are necessary for the skill's primary function of compiling LaTeX documents and processing PDF files.
[DYNAMIC_EXECUTION]: The visual-check action generates a Python script using the fitz library to convert PDF pages into images for visual review. This script is executed dynamically via the command line. This is classified as a low-risk finding because the code is generated from a fixed internal template rather than from untrusted input.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes user-provided external documents (PDF, Markdown, and LaTeX) to generate or review slide content. The lack of explicit boundary markers or instructions to isolate these inputs from the system prompt creates an attack surface for indirect prompt injection.
Ingestion points: Reads .pdf, .md, and .tex files during the Phase 0 analysis and various review actions.
Boundary markers: The skill does not provide instructions for the agent to use delimiters or ignore embedded instructions when reading these external files.
Capability inventory: The skill has access to powerful tools including Bash, Write, Agent, and Edit, which could be abused if an injection is successful.
Sanitization: There is no evidence of sanitization or content filtering for the data processed from external files.

beamer