fix(desktop): keep the @ mention list steady and pick the row you see - #7197

🤖

Summary

Stops the @ mention suggestion list from moving under you, and makes Tab, Enter, and click commit exactly the row you see.

• One request, one list. Opening the list (or changing the typed text) starts a fresh request; you see a loading state while the needed data arrives, then one set of up to 50 choices is installed. Typing more text requests a new set.
• Background changes don't move the list. While the list is open, membership, directory, presence, and ranking updates never replace the identities or labels shown, never reorder the rows, and never move your selected row. The next time the list opens, it uses current evidence.
• What you see is what you choose. Arrow keys select an index in the displayed set; Tab, Enter, and click commit exactly that displayed identity — including rows for two people with the same name, with no global-uniqueness gate. Before, Tab and Enter could race a text-update delay and commit a row from a stale, half-updated list.
• Leaving closes the list. Moving the caret out of the mention, dismissing it, or navigating away abandons the request and the selection; a closed or superseded response can no longer install its results.
• No waiting on a roster that doesn't exist. When the composer has no destination channel ready, the list doesn't wait on a nonexistent roster; a real pending roster still gates readiness.

Scope: display stability only. Permission is still checked live when you choose (#7196) and revalidated at publication; the native editor fixes (highlighting, Enter, caret) are #7323, and the recipient persistence evidence is #7190. The obsolete mention debounce helper that used to paper over the Tab/Enter race is deleted here, together with its test suite.

Part of the mention-chooser stack on the shared #7190 recovery prerequisite: #7190 → #7196 → #7323 → #7197 → #7239 → #7240. This feature stack remains separate from #7191 → #7192. This PR builds directly on #7323.

Related issue

Continues the merged mention-editor work from #7124 (authorize remote mentions at publication) and #7128 (preserve spacing after multi-word mentions). No separate tracking issue for this slice.

Testing

• Unit tests: composerAgentKeyboard.test.mjs, useAgentAddressLockPicker.test.mjs, and mentionAdmissionJourney.test.mjs cover the request lifecycle, stable installation, selection, and closure.
• Browser tests: mention-picker.spec.ts, mentions.spec.ts, remote-owned-mentions.spec.ts, team-mentions.spec.ts, and persistent-agent-audience.spec.ts cover stable rows under background changes, Tab/Enter/click committing the displayed identity (including same-name rows), caret-departure dismissal, and no waiting on a missing destination.
• docs/mention-editor.md documents the stable-completion contract.

Historical capture state (not these heads)

The following original captions/links are retained as historical capture evidence only; none is relabeled as this restack.

Historical visuals — not this rewrite

These immutable captures illustrate the superseded contract (Tab used to refuse ambiguous rows). They are retained for provenance, not evidence of current Tab behavior. Earlier integrated captures are linked below; they do not replace these historical captions.

Historical capture (superseded policy): Tab chooses Scout without an exact choice

At parent revision 6c373187, the example has two recipients named Scout. Tab closes the list and turns the text into a mention for one recipient.

Historical capture (superseded policy): Tab keeps the draft and both choices

At revision e27e770e, Tab leaves @Scout as text and keeps both recipients available.

Historical capture (superseded policy): choosing both recipients keeps them separate

Click the second Scout. Type and @Scout, choose the first Scout, then type hello. The two labels stay separate. One label includes a recipient identifier. The browser test checks that the signed message contains only the two chosen recipient identifiers, in choice order. It does not prove server acceptance or permission to respond.

Integrated stable-picker visuals

Captured integrated f638b9cb — not standalone PR5 and not a capture of this correction. These still illustrate stable rows, action labels and next-open ranking; the earlier correction changed null-destination readiness and fixtures, not those pictured channel states. The duplicate-row image is before Tab, as captioned. No image proves access or delivery.

• Earlier independent-root evidence is historical; current shared-recovery composition validation is recorded below.

Extraction validation update

• This stack builds on the shared fix(desktop): keep your draft editable when an @ mention cannot be sent #7190 recovery baseline, separately from fix(desktop): refresh @ search when channel membership changes #7191 → fix(desktop): show an agent in @ search after you add it to a channel #7192. Chooser, cold-error settlement and ranking production behavior are unchanged by the fixture repairs.
• Desktop lint/format, TypeScript, explicit shared-base file-size checks and E2E builds pass. Targeted send-flow/mention-presentation checks pass. Earlier package 6035/6035 and isolated browser 8/8 remain historical evidence for unchanged semantic inputs, not fresh runs of this composition.
• The fresh composition probe passed 7/8 initially. The Welcome failure was traced to mock parity: create dropped the starter team ID, and add-members omitted the normal membership event. The fixture now preserves team ID, deliberately seeds a same-name collision and delivers that event. Exactly three starter creates prove reuse. The original ambiguous-submit error, retained draft, no-publication and exact-current-starter completion assertions remain.
• Welcome now passes separately on the earliest owning fix(desktop): check access again when you select or pin an @ mention #7196 prefix and final descendant with matching E2E builds. Removing only the fixture membership event fails the roster precondition; restoring it passes. This is not one combined clean eight-test run and does not establish a production freshness dependency on lane B.
• Earlier full browser sweep remains 138/141, with separate separator corrections 2/2 and editor 4/4. No fresh full-suite or all-prefix runtime claim.
• The initial-directory authorization test now holds the actual mock directory response until Loading and no-Quinn assertions finish, rather than spending a one-second delay during navigation. Releasing it passes; withholding release fails eventual visibility. Existing assertions and timeouts are unchanged.
• Independent Welcome fixture review passed; independent held-directory fixture review also passed; earlier helper/shared-base and semantic reviews are retained. DCO passes and this PR is conflict-free. Current-head CI passed, including the repaired initial-directory authorization fixture. The stack remains gated; no ready/merge claim.