pdf-with-mineru

Pass

Audited by Gen Agent Trust Hub on Jul 23, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands (cd, uv run mineru, find) to perform PDF processing and locate resulting files. These commands operate within a specific local directory (~/codes/MinerU) and use user-provided file paths.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by ingesting and acting upon content converted from potentially untrusted PDF files.
  • Ingestion points: The agent reads Markdown content found via find <output_dir> -name "*.md" -type f after conversion.
  • Boundary markers: None; the skill does not specify the use of delimiters or instructions to ignore embedded commands within the extracted Markdown.
  • Capability inventory: The skill possesses shell execution capabilities (uv run mineru), directory navigation, and file reading permissions.
  • Sanitization: There is no evidence of content sanitization, filtering, or validation performed on the extracted text before the agent uses it for tasks like summarization or extraction.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 23, 2026, 06:58 AM
Security Audit — agent-trust-hub — pdf-with-mineru