batch-processor

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill demonstrates the use of standard system tools for parallel file operations, including xargs, find, and GNU parallel. It also provides examples for common command-line utilities such as pandoc, tesseract, and convert (ImageMagick) to perform transformations.
  • [INDIRECT_PROMPT_INJECTION]: The skill defines an attack surface for indirect prompt injection through its document processing workflows.
  • Ingestion points: Files are discovered and read from local directories using find and Python's Path.glob() as shown in the scripts in SKILL.md.
  • Boundary markers: Not present. File contents are processed directly without the use of delimiters or instructions to ignore embedded commands.
  • Capability inventory: The skill facilitates file system read/write operations and the execution of shell commands and external binaries across multiple inputs.
  • Sanitization: There is no evidence of content sanitization or validation of external file data before it is processed or passed to other tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 10:40 AM
Security Audit — agent-trust-hub — batch-processor