The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous instructions and scripts that execute local commands and system utilities. It leverages tools such as qpdf, pdftk, and poppler-utils (including pdftotext and pdfimages) for core PDF operations. These are standard tools for document manipulation and are used according to their intended purposes.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted PDF documents, which serves as an ingestion point for external data. This creates a potential surface for indirect prompt injection if a processed document contains malicious instructions intended to manipulate the agent's behavior during content analysis.
Ingestion points: PDF files are read and processed by scripts/convert_pdf_to_images.py and scripts/extract_form_field_info.py.
Boundary markers: No explicit delimiters are used to separate extracted document content from the agent's internal instruction set.
Capability inventory: The agent has access to file-writing tools via pypdf and is instructed to execute local scripts for document processing.
Sanitization: There is no evidence of sanitization or filtering applied to the text or image data extracted from the documents.
[DYNAMIC_EXECUTION]: The script scripts/fill_fillable_fields.py performs runtime monkeypatching of the pypdf library. It specifically modifies DictionaryObject.get_inherited to fix a known bug in how the library handles selection list fields. This is a targeted functional fix and does not exhibit malicious patterns.
[SAFE]: All referenced Python and Node.js libraries, such as pypdf, pdfplumber, reportlab, and pdf-lib, are standard, widely-used packages from established registries.

pdf