The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute several standard command-line tools for PDF manipulation, including qpdf, pdftotext, pdftk, and pdfimages. These utilities are used for merging, splitting, and extracting content from documents. Additionally, the script scripts/fill_fillable_fields.py modifies the behavior of the pypdf library at runtime to address a known issue with selection list fields, ensuring correct data processing.
[EXTERNAL_DOWNLOADS]: The documentation includes instructions for installing well-known and trusted Python libraries such as pytesseract, pdf2image, and pypdfium2 from standard package registries. These libraries are used for advanced features like OCR and high-fidelity PDF rendering.
[PROMPT_INJECTION]: As a tool designed to ingest and process external PDF files, the skill has an inherent exposure to indirect prompt injection. Malicious instructions or content embedded within a PDF's text or metadata could potentially influence the agent's logic during document analysis or form-filling workflows.
Ingestion points: The skill reads document content and form metadata using pypdf and pdfplumber from user-supplied PDF files.
Boundary markers: No explicit delimiter markers are used in the scripts for handling extracted content, although the instructions require the agent to perform a multi-step visual validation process.
Capability inventory: The skill utilizes shell command execution, file system writes for document generation, and PDF annotation capabilities.
Sanitization: The form-filling scripts include logic to validate provided field IDs and values against the document's internal metadata before committing changes.

pdf