cardputer-buddy

Pass

Audited by Gen Agent Trust Hub on Nov 3, 2026

Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructions involve running local Python scripts such as install_apps.py, push.py, and tail_serial.py via the command line. These scripts are used to manage a physical device and are consistent with the skill's primary development purpose.\n- [DYNAMIC_EXECUTION]: The skill includes a repl_run.py utility that allows the agent to execute arbitrary MicroPython commands on the connected hardware device. This functionality enables runtime control of the external device's environment.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from a hardware device's serial output, creating a potential surface for indirect injection.\n
  • Ingestion points: Serial logs and device output are read into the agent context using buddy/scripts/tail_serial.py.\n
  • Boundary markers: The instructions do not specify any delimiters or safety markers to isolate the device output from the agent's instructions.\n
  • Capability inventory: The skill allows for local shell command execution and arbitrary MicroPython REPL execution on the target hardware.\n
  • Sanitization: There is no description of output filtering or validation for the data received from the device before it is processed.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 3, 2026, 10:57 AM