Skills
skills/pudap/skills/puda-machines/Gen Agent Trust Hub

puda-machines

Pass

Audited by Gen Agent Trust Hub on Apr 16, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The agent is instructed to run puda CLI commands and uv package commands to explore machine capabilities and manage edge services. These commands interact with local system resources and hardware serial ports (e.g., /dev/ttyUSB0) required for machine operation.\n- [PROMPT_INJECTION]: The opentrons machine reference describes reading data from external CSV files using read_csv_file. This is an indirect prompt injection surface where untrusted input could influence generated protocols.\n
  • Ingestion points: read_csv_file and read_csv in references/opentrons-machine.md\n
  • Boundary markers: Missing explicit delimiters or instructions to ignore embedded commands in the CSV data\n
  • Capability inventory: Includes shell execution and generation of runnable Python protocol code\n
  • Sanitization: No explicit sanitization or validation of CSV content is defined.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 16, 2026, 08:41 AM