invoice-organizer

Pass

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses shell commands like find, mkdir, cp, and mv to scan, organize, and rename files on the local filesystem.
  • [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by processing untrusted data.
  • Ingestion points: Reads text content and metadata from external PDF, JPG, and PNG files (invoices/receipts).
  • Boundary markers: Absent. There are no instructions to the agent to treat extracted text as untrusted or to use delimiters when processing file content.
  • Capability inventory: The skill has access to bash execution for file operations (find, mkdir, cp, mv), the ability to read arbitrary file contents, and the ability to write new files (CSV reports).
  • Sanitization: Absent. The instructions do not specify sanitization or validation of the extracted vendor names, dates, or descriptions before using them in command-line operations or CSV generation.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 6, 2026, 01:57 PM