Thanks @svabi79 — this is a careful follow-up to #3401, and the structure is right: everything stays behind HAVE_HIDAPI, the new glyph writers preserve the shared indicator bits with the same masks as the digit writers (0xF8/0xF0 main, 0x0F/0x1F small), settings go through AppSettings, and the TX-watts updates were correctly placed inside the already-amp-gated meter lambdas, so the exciter/amp alternating-writer race from #2927 doesn't reappear on the LCD. The E1/E2 ↔ TMate2EncoderAction2/1 swap is well-commented and its hardcoded fallbacks match the canonical defaults in tmate2EncoderDefaultAction().

Two real display bugs to fix before merge:

1. Missing 'D' glyph — "AGC MED" renders as "AGC ME " (HidEncoderManager.cpp, tmate2MainGlyph)
The glyph table has no case for 'D'/'d', so it falls through to blank. ToggleAgc's labels include "AGC MED", which will display as AGC ME. Please add a 'D' glyph (the usual 7-segment lowercase-d: segments B+C+D+E+G) — worth a quick hardware check since you have the unit on the bench.

2. 10 kHz step overlay truncates to "STEP 1000" (MainWindow_Controllers.cpp, updateTMate2Display)
Speed formats QStringLiteral("STEP %1").arg(mainVal, 4, …). RxApplet::m_stepSizes goes up to 10000, so the string becomes "STEP 10000" (10 chars) and tmate2WriteMainDisplayText keeps only text.left(9) → the display reads STEP 1000 for the 10 kHz step — the previous numeric path handled this correctly, so this is a regression. Suggest a shorter label ("STP %1") or formatting ≥1000 Hz steps as "STEP 10k" (the K glyph exists).

Minor:

• isWpm is now unused in setTMate2OverlayIndicators() after applyModeSegs(QString()) replaced the CW-forcing — this will trip -Wunused-variable under the project's -Wall -Wextra. Just drop the variable.
• formatTMate2PowerSmallText() doesn't guard non-finite input — std::lround(NaN) is unspecified. powerBars() checks std::isfinite; the small-text formatter takes the same meter-derived float, so a one-line if (!std::isfinite(watts)) watts = 0.0f; would make them symmetric.
• TMate2Overlay::Shift is added but never triggered — nothing calls triggerTMate2Overlay(TMate2Overlay::Shift, …), and its text case currently shares the "STEP" label with Speed, which would read wrong if it's ever wired up. Fine if it's groundwork for a follow-up; if so, a // not yet triggered note (or giving it its own label now) would prevent surprise later.

No scope concerns — the five files match the stated feature, no new settings keys, and the setTMate2Indicators signature extension uses default-valued params so any other caller stays source-compatible. With the two display bugs fixed this looks good.

_{🤖 aethersdr-agent · cost: $17.0298 · model: claude-fable-5}

Follow-up to #3401 reviewed cleanly. All five issues from the bot review resolved in dd8ba6f0 and hardware-confirmed by the author on the TMate 2 bench:

• 'D' glyph ({ 0x06, 0x0E }) correctly encodes B+C+D+E+G; "AGC MED" now renders in full.
• STEP overflow resolved by switching to "STP %1".arg(mainVal, 5) → "STP 10000" is exactly 9 characters at the 10 kHz preset.
• isWpm dropped; non-finite guard added to formatTMate2PowerSmallText symmetric with powerBars(); TMate2Overlay::Shift given its own "SHFT" label with a // not yet triggered note.

Code placement: The PR body's files table lists MainWindow.cpp for the TX power wiring, but the actual diff is in MainWindow_Wiring.cpp (inside wireMeters()) — the correct TU per the #3351 decomposition. MainWindow.cpp is not modified. All overlay/indicator logic is in MainWindow_Controllers.cpp. Fully aligned with the refactored standard.

TX bargraph: Three power update paths (exciter FWD, PGxl forwarded, transmitPowerChanged) now all write m_tmate2TxWatts and call both updateTMate2Display() and updateTMate2Indicators(). The txPowerFullScaleWatts re-evaluates live via hasAmplifier() && ampOperate() at each updateTMate2Indicators() call, so the 100W → 2000W rescale tracks amp state without a stale capture. The amp/exciter alternating-writer race from #2927 does not re-emerge on the LCD path.

Encoder labels / RIT–XIT double-function: E1/E2 settings-key swap is documented inline and consistent with tmate2EncoderDefaultAction(). RIT/XIT segments lit on encoder assignment (even when offset is zero) is the right UX for showing knob function without an active rit/xit offset.

No new settings keys, no DAX, no session changes. CI 6/6 green. Approved.