The handoff document

scope: generic · severity: technique · confidence: proven · subsystem: method

A bring-up is a relay. Sessions end mid-experiment, context is lost, and the next reader — often an agent with no memory of any of it — needs to pick up without re-deriving.

A handoff that works has five parts:

State right now. What is flashed, what is running, what physical state the device is in. First, because it is what a wrong assumption costs most.
What was proven this session. Observations with evidence. Include the refutations.
What was NOT proven. Explicitly. This is the part everyone skips and the part that saves the most time: it stops the next session re-running an inconclusive experiment and reaching the same non-conclusion.
The next concrete step. Not “investigate suspend” — the actual command, with the actual arm, and what result would mean what.
Anything currently unsafe. A half-flashed slot, a disarmed watchdog, an experiment someone else has running.

Supersede rather than accumulate. Twenty handoffs where nineteen are stale is worse than one that is current, because a reader cannot tell which is which. Name the file so the newest is obvious, and say in it which document it replaces.