markdown-converter

Pass

Audited by Gen Agent Trust Hub on Apr 12, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the uvx tool to execute markitdown, a document conversion utility from Microsoft. It also executes a local feedback recording script (python3 utilities/skill-builder/scripts/record_skill_feedback.py) to log usage decisions.
  • [DATA_EXFILTRATION]: Includes explicit constraints to redact secrets, credentials, and sensitive content by default when presenting sample outputs, reducing the risk of accidental data exposure.
  • [PROMPT_INJECTION]: The skill processes untrusted external document formats (PDF, DOCX, HTML) and URLs. This represents an ingestion surface for indirect prompt injection where malicious instructions could be embedded in source documents.
  • Ingestion points: User-provided file paths, URLs, or stdin content (SKILL.md).
  • Boundary markers: None explicitly defined for the content being converted.
  • Capability inventory: Subprocess execution via uvx markitdown and python3 (SKILL.md).
  • Sanitization: Instructions mandate the redaction of secrets and sensitive content in sample outputs.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 12, 2026, 01:13 PM
Security Audit — agent-trust-hub — markdown-converter