The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to extract text and structured data from user-provided PDF files. This extracted content is then interpolated into prompts for AI models to perform semantic analysis.
Ingestion points: PDF files are processed using libraries like unpdf, pdfplumber, and pypdf as described in SKILL.md, PDF-PROCESSING.md, and references/1-ai-extraction-patterns.md.
Boundary markers: The implementation examples (notably in references/1-ai-extraction-patterns.md) do not demonstrate the use of strict delimiters or instructions to ignore potential commands embedded within the extracted PDF text when passing it to an LLM.
Capability inventory: The skill utilizes tools with significant capabilities, including shell execution for structural repairs (qpdf, poppler-utils), file system writes for document generation, and network access via headless browsers (Puppeteer) for HTML-to-PDF conversion.
Sanitization: No explicit sanitization or filtering of the extracted text is shown before it is included in the LLM prompt, which is an inherent risk in document processing workflows.
[DYNAMIC_EXECUTION]: The script scripts/fill_fillable_fields.py performs a runtime monkeypatch of the pypdf library to fix a specific bug in its handling of selection list fields.
Evidence: The function monkeypatch_pydpf_method redefines DictionaryObject.get_inherited at runtime. While this is documented as a necessary workaround for a library limitation, runtime modification of third-party code is a dynamic execution pattern that should be monitored.

pdf-pro