The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted PDF content, creating a surface for indirect prompt injection where adversarial text in a document could attempt to influence the agent's behavior.
Ingestion points: Functions such as pdf_pages, pdf_scan, and pdf_grep in kernel.py read text and images directly from user-provided PDF files.
Boundary markers: The skill does not currently implement specific delimiters or instructions to ignore embedded commands within the extracted PDF text.
Capability inventory: The skill's capabilities are restricted to local PDF parsing and image rendering; it does not have access to the network or sensitive system credentials.
Sanitization: No filtering or sanitization is applied to the extracted text before it is returned to the agent.
[EXTERNAL_DOWNLOADS]: The skill relies on external Python libraries for PDF processing and image rendering.
Evidence: The setup instructions in SKILL.md recommend installing pypdfium2 and pillow via pip. These are well-known, legitimate packages from the official Python Package Index.
[COMMAND_EXECUTION]: The skill uses dynamic execution to initialize its helper functions within the agent's environment.
Evidence: The skill instructions use exec(open(...).read()) to load the kernel.py file. This is a standard method for loading this skill's logic into the active session.

pdf-explore