Test point

A pad whose only job is to be touchable. Nothing connects to it and nothing is built from it. It exists so that someone holding a probe, or a bed-of-nails fixture on a production line, can reach a net that is otherwise buried under a component or inside the board.

flowchart TB
    RG["regulator"] --> RAIL["3V3 rail"]
    RAIL --> LOAD["the parts it feeds"]
    RAIL --> TP["test point<br/>a pad, connected to nothing else"]
    TP -.->|"a probe, or a<br/>bed-of-nails pin"| M["is this rail<br/>actually at 3.3V?"]

When a board does not work, the first question anybody asks is which rails came up. A test point answers it. Without a reachable point, you have to find a component leg to balance a probe on. That is slow on a bench and impossible on a production fixture.

So test-point-coverage asks whether every rail carries one. The rule only fires on a board that places test points somewhere, since a design with no test-point convention is not wrong, it just has a different one.

For a software reader, a test point is a metrics endpoint, and the rule reads as “critical paths must emit telemetry”.

Chapter 13 of the course teaches it in full, along with the other ways a probe reaches a net and why in-circuit test needs a test point on both ends of a part.