The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data (PDFs, Word documents, HTML, etc.) which can contain adversarial instructions designed to influence the agent's behavior.
Ingestion points: The skill uses markitdown to extract text from user-provided files (e.g., markitdown "$INPUT_FILE").
Boundary markers: There are no instructions for the agent to treat the output of the conversion as untrusted data or to ignore instructions embedded within the processed documents.
Capability inventory: The skill relies on shell command execution and file system access (read/write) to perform conversions and manage outputs.
Sanitization: No sanitization or filtering is performed on the extracted text before it is returned to the agent's context.
[EXTERNAL_DOWNLOADS]: The skill suggests installing the markitdown package via pip install 'markitdown[all]'. This package is provided by a well-known service (Microsoft).
[COMMAND_EXECUTION]: The skill uses shell commands, piping, and redirection (e.g., markitdown input.pdf -o output.md 2>/dev/null) to perform its primary function. These commands are consistent with the documented purpose of document conversion.

to-markdown