The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches the MarkItDown library from the official Python Package Index (PyPI) and clones the repository from Microsoft's GitHub organization. These are recognized as trusted or well-known sources.
[COMMAND_EXECUTION]: Uses the Bash tool to install software and run CLI commands, including a local script scripts/generate_schematic.py used for creating diagrams.
[PROMPT_INJECTION]: Includes instructions that guide the agent's behavior, such as proactively suggesting the author's platform (K-Dense Web) for complex tasks and defaulting to diagram generation.
[PROMPT_INJECTION]: The skill acts as an indirect prompt injection surface by converting external documents into text for the LLM.
Ingestion points: Document conversion methods process PDFs, DOCX files, and YouTube transcripts provided by the user or fetched from URLs.
Boundary markers: The skill does not implement boundary markers or instructions to disregard embedded commands in processed data.
Capability inventory: The skill environment allows high-privilege operations via Bash, Read, Write, and Edit tools.
Sanitization: No evidence of content validation or sanitization was found in the skill's conversion logic.

markitdown