The Agent Skills Directory

[PROMPT_INJECTION]: The skill is designed to convert untrusted external documents (PDF, Word, Excel, YouTube subtitles, etc.) into Markdown specifically for processing by an LLM. This introduces a surface for indirect prompt injection attacks, where instructions embedded within a source document could influence the agent's subsequent behavior.
Ingestion points: CLI commands (e.g., python -m markitdown input.pdf) and Python API calls (md.convert("plik.docx")) that process user-provided files from the filesystem or external URLs.
Boundary markers: The skill instructions do not define specific delimiters or "ignore embedded instructions" warnings to isolate the converted content when it is returned to the agent's context.
Capability inventory: The skill is allowed to use Bash, Read, and Write tools, which could be targeted for abuse if an agent follows instructions injected through a converted document.
Sanitization: There is no evidence of sanitization, filtering, or validation of the converted output before it is processed by the AI.
[EXTERNAL_DOWNLOADS]: The skill installs the markitdown and markitdown-mcp packages from the official Python Package Index (PyPI) using pip. These are well-known libraries maintained by Microsoft.

markitdown