The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of the @llamaindex/liteparse package from the NPM registry as noted in the setup instructions in SKILL.md. Additionally, the scripts/search.py helper script utilizes the uv tool to dynamically manage and download dependencies including bm25s and aiofiles from PyPI.
[COMMAND_EXECUTION]: The skill instructions frequently utilize shell commands to perform document processing and file management.
Specifically, it executes the lit CLI tool for document parsing and screenshotting.
It invokes the bundled scripts/search.py Python script to perform ranked searches.
It uses standard Unix utilities such as grep, sed, head, and wc to manipulate extracted text files.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external, potentially untrusted documents (PDF, DOCX, PPTX, XLSX, and images). This creates a vulnerability surface where malicious instructions embedded in documents could influence the agent's behavior.
Ingestion points: Document files are read via lit parse in SKILL.md and scripts/search.py.
Boundary markers: None identified. Extracted content is piped directly into the agent's context using tools like grep or printed by the search helper.
Capability inventory: The skill has access to shell execution (lit, python, grep, sed) and file writing (-o /tmp/doc.txt).
Sanitization: There is no evidence of sanitization, filtering, or escaping of the text extracted from documents before it is presented to the agent.

liteparse