document-extractor
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to convert external, potentially untrusted data sources (such as PDFs, Office documents, and YouTube transcripts) into Markdown for agent consumption. This ingestion of external content presents a surface for indirect prompt injection if the source files contain hidden instructions intended to influence the agent's behavior.
- Ingestion points: Local files and YouTube URLs.
- Boundary markers: No explicit delimiters or safety warnings are included in the conversion commands provided to the agent.
- Capability inventory: The skill utilizes the
markitdownCLI and Python API to perform file reads and format conversions. - Sanitization: The skill relies on the underlying
markitdownlibrary for data processing and does not implement additional sanitization of the extracted text content. - [EXTERNAL_DOWNLOADS]: The skill provides instructions to install the
markitdownpackage from standard registries and references documentation and source code hosted on Microsoft's official GitHub repository. - [COMMAND_EXECUTION]: The skill utilizes the
markitdownCLI for document conversion, plugin listing, and YouTube transcript extraction. It also includes examples for using the tool's Python API.
Audit Metadata