The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on executing the uvx markitdown command to process local and remote files. This is the primary function of the skill and allows the agent to transform binary or structured documents into text format.
[EXTERNAL_DOWNLOADS]: The skill uses uvx, which automatically downloads the markitdown package and its dependencies from the Python Package Index (PyPI) during the first execution or if not cached. This is standard behavior for the uv toolchain when running packages from well-known registries.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and convert untrusted external data (PDFs, Word docs, HTML, etc.) into Markdown for further processing by the agent. This introduces a vulnerability surface where instructions hidden within these documents could be interpreted by the LLM as system-level commands.
Ingestion points: Any document or file path provided to the markitdown command (SKILL.md).
Boundary markers: None explicitly defined in the skill instructions to separate the converted text from the agent's instructions.
Capability inventory: Execution of shell commands via uvx (SKILL.md).
Sanitization: The skill relies on the underlying markitdown tool to extract text, but does not implement additional filtering for malicious prompt content.

markdown-converter