skills/bulls-cows/skills/yy-read-pdf/Gen Agent Trust Hub

yy-read-pdf

Pass

Audited by Gen Agent Trust Hub on May 20, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to perform shell operations, including installing Python dependencies via pip and executing a local script (scripts/read_pdf.py) to process user-provided files.
  • [EXTERNAL_DOWNLOADS]: The skill requests the installation of the pymupdf package from the public PyPI registry. This is a standard requirement for the skill's PDF parsing functionality.
  • [INDIRECT_PROMPT_INJECTION]: As a tool that ingests and extracts text from external documents (PDFs), it is susceptible to indirect prompt injection. Malicious instructions embedded within a PDF could attempt to hijack the agent's context once the text is processed.
  • Ingestion points: scripts/read_pdf.py reads local PDF files.
  • Boundary markers: None explicitly defined in the prompts or script to isolate extracted text.
  • Capability inventory: File read access via fitz.open and command execution via the agent's terminal.
  • Sanitization: The script performs no sanitization or escaping of the extracted text before it is returned to the agent.
  • [COMMAND_EXECUTION]: There is a potential risk of shell command injection if the agent naively interpolates user-provided file paths into the python command string without proper escaping or validation.
Audit Metadata
Risk Level
SAFE
Analyzed
May 20, 2026, 04:40 AM
Security Audit — agent-trust-hub — yy-read-pdf