The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands for environment setup, specifically using uv pip install for Python packages and system package managers like apt or brew for OCR tools.
[EXTERNAL_DOWNLOADS]: The skill downloads software and configuration from reputable sources, including official Python packages (markitdown, trafilatura) and browser binaries for Playwright.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it processes untrusted content from external URLs and documents.
Ingestion points: Web URLs and various local file types (PDF, Office, Images) as specified in SKILL.md and fetch.md.
Boundary markers: Absent; there are no specific instructions to use delimiters or ignore instructions within the converted content.
Capability inventory: The skill can install arbitrary packages, run Python scripts, execute shell commands, and write files to the system.
Sanitization: Content is formatted into Markdown using library-standard conversion logic without explicit sanitization or filtering of potential injection patterns.

to-markdown