summarize

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • EXTERNAL_DOWNLOADS (LOW): The skill uses uvx to fetch and execute the markitdown package from PyPI. While Microsoft is a trusted organization, dynamic package fetching is an external dependency risk.
  • COMMAND_EXECUTION (LOW): The skill executes shell commands (uvx, node, pi) to perform document conversion and LLM interaction.
  • PROMPT_INJECTION (LOW): Category 8 (Indirect Prompt Injection): The skill ingests untrusted data from URLs or local files and feeds the output into an LLM. Ingestion points: Path and URL arguments passed to markitdown. Boundary markers: Absent. Capability inventory: File reading and LLM summarization via the pi command. Sanitization: None observed in the provided instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:17 PM