reading-payment-statement

Pass

Audited by Gen Agent Trust Hub on Feb 25, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses a custom CLI utility shinkoku to perform text extraction and image conversion on PDF files. These commands (shinkoku pdf extract-text and shinkoku pdf to-image) are used for the primary purpose of document digitizing.
  • [PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it processes untrusted external documents.
  • Ingestion points: File contents processed via shinkoku and the Read tool in SKILL.md.
  • Boundary markers: The skill defines output boundaries but lacks specific instructions to disregard instructions found within the input images or PDFs.
  • Capability inventory: Execution of shell commands (shinkoku) and file system access.
  • Sanitization: No input sanitization or filtering of extracted text is mentioned before the data is processed by the model.
  • [SAFE]: The skill implements a security-positive 'dual verification' pattern, requiring two independent agents to agree on extracted values, which mitigates risks of data hallucination or manipulation.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 25, 2026, 05:13 AM