The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes external PDF files, creating a surface for indirect prompt injection where malicious instructions could be embedded in document text or form metadata.
Ingestion points: Data enters the agent's context through text and metadata extraction in scripts/extract_form_structure.py, scripts/extract_form_field_info.py, and library code snippets in SKILL.md.
Boundary markers: No specific delimiters or safety instructions are used to separate extracted PDF content from the agent's core instructions.
Capability inventory: The skill has broad capabilities including file system access (read/write) and execution of external command-line utilities.
Sanitization: Extracted content is not sanitized or escaped before being presented to the agent.
[DYNAMIC_EXECUTION]: The script scripts/fill_fillable_fields.py performs a runtime monkeypatch of the pypdf library. It modifies DictionaryObject.get_inherited at runtime to adjust handling of form field attributes, which constitutes dynamic code modification.
[COMMAND_EXECUTION]: Documentation in SKILL.md and forms.md suggests the use of various CLI tools such as qpdf, pdftotext, and magick for PDF and image manipulation. These are standard operations for the skill's intended purpose.
[EXTERNAL_DOWNLOADS]: The skill references several well-known technology libraries such as pypdf, pdfplumber, and reportlab. These are recognized as well-known, trusted services and are documented neutrally as functional requirements.

pdf