The one gesture

Hold Ctrl+Win, speak, release. The text lands at the caret of whatever has focus and your clipboard is put back. That is the product; everything below is a variation on it.

Double-tap Ctrl+Win for free mode — it keeps recording until you double-tap again, then pastes the finished note at the caret, leaves it on the clipboard, and saves it under .deixis\notes\. A saved note with pictures in it also has Copy with images in Captures (Windows): the same text, plus the note's own pictures inlined as HTML, for a document or any editor that takes pasted pictures. Copy stays the text with file paths, which is what a terminal agent opens.

Hold the chord and the mode palette appears above the listening pill: one chip per mode, with its letter. Eight ship — S screen grab, N hands-free note, M meeting, V screen recording, F free mode, D dictionary, T translate and C Claude usage — and the authoritative list is the ACTIONS table in core/src/hotkey.rs, not this paragraph. They are defaults: [mode_keys] in your config reassigns them, and a mode the platform cannot do is dropped rather than shown.

The letters stay live for 5 s after the chord engages (ACTION_WINDOW_MS) and the palette fades exactly when they stop working — later than that, S is a letter again. Press the letter while the chord is held, or click the chip. DEIXIS_ACTION_KEYS=off hands the letters back to typing, custom buttons included.