Once your AI assistant is connected to both probe and trace (in Claude Code, /mcp shows both as connected), you can ask questions in plain language. The assistant picks the right tools itself; you do not need to name tools or know how the data is organised.
The best test is a board you have already reviewed, where you know what the problems were. The point of these questions is to see whether Probe finds them, so start broad and then push on whatever comes back.
Start here
Review this board and give me your top concerns, worst first.
Is there anything here that would stop this board being manufactured?
Do the BOM and the schematic agree on every part?
Is there anything that could damage a component at power-up?
Then press on the answers
Show me the evidence for that finding. How do you know?
What have you not been able to check, and why?
Which of these would you fix before the next spin, and which can wait?
Interconnect (two or more boards)
Check the connections between these two boards. Do the pin assignments and signals match across the connector?
Are the power rails compatible across the interconnect?
Push back on anything that looks thin, and ask the assistant to justify a finding. It can follow its own reasoning further and will tell you when the evidence runs out.