The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external PDF files, which presents an attack surface where malicious instructions embedded within documents could potentially influence the agent's behavior.
Ingestion points: Files are read in briefs/process.md using pypdf and pdfplumber.
Boundary markers: The skill does not define specific delimiters or instructions for the agent to ignore embedded content during extraction.
Capability inventory: The skill has access to shell execution (tectonic, libreoffice) and file system write operations.
Sanitization: There is no explicit sanitization or safety validation applied to text extracted from processed PDFs.
[EXTERNAL_DOWNLOADS]: The scripts/ensure_fonts.py script automatically fetches font assets from Google's official GitHub repository (raw.githubusercontent.com/google/fonts). This is a legitimate operation for acquiring Open Font License (OFL) resources needed for CJK document rendering.
[COMMAND_EXECUTION]: The skill facilitates its core functionality by executing several command-line tools, including tectonic for LaTeX compilation and libreoffice for document format conversion.
[DYNAMIC_EXECUTION]: The font helper script utilizes the fontTools library to dynamically generate static font instances from variable font files at runtime. The workflow in briefs/process.md also involves spawning subprocesses for OCR and document conversion tasks.

pdf-production