The Agent Skills Directory

[SAFE]: The skill implements local document parsing using the 'liteparse' library (version 2.0.0). All documented workflows focus on local processing without cloud dependencies.
[EXTERNAL_DOWNLOADS]: The skill references standard installation procedures for well-known and trusted tools:
Python package 'liteparse' from PyPI.
Node.js package '@llamaindex/liteparse' from npm.
System dependencies including LibreOffice, ImageMagick, and Tesseract via standard package managers (brew, apt-get, choco).
[COMMAND_EXECUTION]: The CLI tool 'lit' and the 'batch_parse_dir.py' script perform legitimate file system operations (reading documents, writing text/JSON outputs) and call system binaries (LibreOffice's 'soffice', ImageMagick's 'convert') required for its primary purpose of document conversion and parsing.
[DATA_EXFILTRATION]: While the skill mentions an optional --ocr-server-url for HTTP-based OCR, this is presented as a user-configurable feature for custom infrastructure rather than a hardcoded exfiltration path. Standard usage remains local.
[PROMPT_INJECTION]: No prompt injection or behavior override patterns were found in the instructions or metadata.
[INDIRECT_PROMPT_INJECTION]: The skill has a data ingestion surface (parsing untrusted PDFs/Office files). It is classified as safe because it returns raw text/JSON data for further processing rather than executing instructions contained within the documents, and it includes clear agent operating procedures for validating results.

liteparse