skills/zhihaoairobotic/clawphd/pdf2md/Gen Agent Trust Hub

pdf2md

Pass

Audited by Gen Agent Trust Hub on May 15, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted PDF files to generate Markdown content for the agent, creating a surface for indirect prompt injection. Malicious instructions inside a PDF could be parsed and followed by the agent in subsequent steps. • Ingestion points: pdf_to_markdown tool (SKILL.md). • Boundary markers: Absent. • Capability inventory: File system access (read/write), system command execution, and network API calls (fal.ai, OpenRouter/Gemini). • Sanitization: Absent.
  • [COMMAND_EXECUTION]: The skill invokes system CLI tools like mutool, pdf2svg, and mineru (magic-pdf) for PDF parsing and image extraction tasks.
  • [EXTERNAL_DOWNLOADS]: The skill requires the clawphd-ai Python package and accesses remote AI APIs (fal.ai, OpenRouter, Gemini) for figure reconstruction tasks.
Audit Metadata
Risk Level
SAFE
Analyzed
May 15, 2026, 07:37 AM
Security Audit — agent-trust-hub — pdf2md