skills/kyosakuyo/skills/pdf-set/Gen Agent Trust Hub

pdf-set

Pass

Audited by Gen Agent Trust Hub on Mar 1, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires the installation of the pypdfium2 and google-genai packages from the official Python Package Index (PyPI). These are standard libraries for PDF rendering and accessing the Google Gemini API.
  • [CREDENTIALS_UNSAFE]: The script scripts/ocr.py is designed to read API credentials (api_key, api_endpoint) from a local file named secrets.txt. This is a standard pattern for local tools to manage user-provided API keys. The skill does not contain hardcoded secrets or evidence of exfiltrating these credentials to unauthorized third parties.
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute several local Python scripts (e.g., ocr.py, convert_pdf_to_images.py) to perform its tasks. These commands are limited to processing local files and interacting with the generative AI API as configured by the user.
  • [PROMPT_INJECTION]: The skill contains specific instructions for the AI agent's behavior (e.g., instructions in references/OCR.md to stop output and wait for the user). These are workflow management guidelines designed to coordinate task execution and do not attempt to bypass core safety guardrails or extract system prompts.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 1, 2026, 06:47 PM