markitdown-cli
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the installation of the
markitdownpackage and its optional components (such aspdf,docx,pptx,xlsx,az-doc-intel, andyoutube-transcription) viapip installas seen inSKILL.mdandexamples/basic/first-commands.md. These downloads originate from PyPI. - [INDIRECT_PROMPT_INJECTION]: The skill is configured to ingest and process data from untrusted external sources.
- Ingestion points: The tool accepts local file paths, standard input stream (stdin), and remote HTTP/HTTPS URLs as input for conversion, as detailed in
SKILL.mdand the batch scripts inexamples/advanced/workflows.md. - Boundary markers: The instructions do not mandate the use of delimiters or specific safety markers when the agent processes content retrieved from these sources.
- Capability inventory: The skill enables file system reads/writes and network requests (via the tool's URI conversion feature). It also includes shell scripts that automate file processing across directories.
- Sanitization: The skill lacks explicit sanitization or validation steps for the content or metadata of the files and URLs it processes.
Audit Metadata