# ✅ `screen_id.py` filed the mission briefing as `menu` — a successful route scored as a failure

**Confidence: ✅ CONFIRMED and FIXED.** Measured 2026-08-26.

## Symptom

`launch_mission.sh`'s nav ran to completion and `wait_screen.sh readyroom 300`
reported:

```
NO readyroom within 300s (last seen: menu)
NEVER REACHED READY ROOM
```

The obvious reading — and the one taken at first — is that the presses after the
main menu never landed. That reading was wrong twice over:

1. Pad input *was* landing. With the game's own cursor counter as the oracle, two
   `dpad down` presses moved it `24 → 26 → 28`.
2. The run was not on the main menu. It was on the **mission briefing map**
   ("Glasner Training Area", Ⓐ:Continue / Ⓑ:Back / START:Skip) — three screens
   past the menu, i.e. `LOAD GAME → slot 01 → YES` had all worked.

![briefing misread as menu](captures/briefing-misclassified-as-menu.png)

## Cause

The briefing is **cyan**; the menu rule tests for *dark and blue* and the
briefing satisfies every clause of it — `b - r > 30`, `r < 45`, `white < 0.075`.
With no earlier rule claiming it (the ready-room rule needs `b > 110`, the flight
rule needs green), it fell through and was labelled `menu`.

Ironically the file's own docstring already recorded the distinguishing number:
it notes the briefing as "cyan rather than blue at (4.6, 47.7, 54.5)" — as an
aside justifying the ready-room threshold, never as a class of its own.

## Fix

Cyan has blue and green nearly equal; the menu's blue runs far ahead of its green:

| screen | measured means | `b - g` |
|---|---|---|
| briefing | (1.4, 45.2, 49.6), (5.5, 49.1, 54.3), (4.6, 47.7, 54.5) | ~5 |
| main menu | (25, 38, 70), (13, 26, 59) | ~32 |

So `r < 20 and g > 30 and b - g < 20` → `briefing`, placed above the menu rule.
The `r < 20` floor keeps the title screen (52.8, 66.1, 74.5, `b - g` 8.4) out;
the ready room (`b - g` ~97) and flight (green rule) are already claimed above.

Verified against **every** signature the file documents — the two menu variants,
title, ready room, flight, and three briefing measurements: 8/8 classify as
intended, no regressions. The real capture above now reads `briefing`.

## Why it mattered

This is the expensive failure mode, not a cosmetic one: a route that *worked*
was reported as broken, and the conclusion drawn from it — "the presses are not
landing" — would have sent the next iteration to debug an input path that was
fine. Compare the same file's earlier note about a run that "sat on a plainly
visible MAIN MENU for 300 s": same class of bug, second occurrence.