fix(cli): monitor prompt handoff no longer races repaint teardown

[teknium1]

CLI monitor repaint requests no longer race the nested application's event-loop teardown.

• Route both immediate modal paints and spinner-thread monitor refreshes through one monitor repaint entry point on the existing UI-loop scheduler.
• Capture the application once, so monitor detachment cannot change the target between checks.
• Extend the existing prompt-handoff invariant with event-synchronized teardown for both callers; retain all five prompt types, ordinary-paint behavior, dock suppression, and secret draft restoration. No added sleep or timeout increase.

Root cause: prompt_toolkit's thread-safe invalidate() reads app.loop multiple times; a queued modal frame can finish the nested application between those reads and leave a worker calling call_soon_threadsafe(loop=None).

This was found in the first-attempt output of CI run 34263993728 for #105960 (retry passed), not a reported user-machine incident. This PR is separate from the UI work and does not modify it.

Validation

Check	Result
Event-synchronized current-main reproduction	Both modal and spinner-tick cases fail with RuntimeError: no running event loop; unchanged secret test passes
Fixed invariant stress, canonical runner, retries disabled	10 complete file runs, 3 passing cases each, zero failures
Live repro: native Linux PTY, real HermesCLI layout, nested in_terminal() monitor and secret callback	Controlled before: exact RuntimeError; controlled after and uninstrumented fixed control: no errors, monitor stopped, original draft and cursor restored
Canonical tests/cli/ + tests/hermes_cli/, retries disabled	987 files: 10,619 passed, 1 failed, 137 skipped. Sole failure is the separately addressed root-home fixture in #105920; no duplicate fix here
Final modified test after explicit ordinary-frame synchronization	Canonical 10-run stress passed
Static checks	Ruff, Windows-footgun scan, compat-pointer check, diff whitespace passed

Native evidence uses a disposable credential-free profile and the shipped CLI widget/callback implementations; it controls worker scheduling to reproduce the race, not a live model request or historical CI scheduler replay. An earlier interrupted directory run also hit test_update_success_when_head_moves; the completed directory run passed that file.

Infographic

[github-actions]

૮ >ﻌ< ა ci review

_{ran on 497ef9e — fix(cli): keep monitor repaints safe during prompt handoff}

⚠️ Warnings

OSV vulnerability scan · View job

28 known vulnerabilities found in pinned dependencies.

• CVE-2026-67213 in website/package-lock.json
• CVE-2026-82417 in scripts/whatsapp-bridge/package-lock.json
• CVE-2026-82417 in website/package-lock.json
• CVE-2026-75931 in package-lock.json
• CVE-2026-75931 in website/package-lock.json
• CVE-2026-83610 in package-lock.json
• CVE-2026-83610 in package-lock.json
• CVE-2026-71554 in uv.lock
• CVE-2026-73088 in package-lock.json
• CVE-2026-73088 in website/package-lock.json
• GHSA-8423-8fgw-73vq in uv.lock
• CVE-2026-70608 in package-lock.json
• CVE-2026-73089 in package-lock.json
• CVE-2026-73089 in website/package-lock.json
• CVE-2026-75975 in package-lock.json
• CVE-2026-75975 in website/package-lock.json
• CVE-2026-75899 in package-lock.json
• CVE-2026-75899 in website/package-lock.json
• CVE-2026-84371 in package-lock.json
• CVE-2026-56876 in package-lock.json

How to fix:

Review the findings in the Security tab. Update the affected dependencies if a patched version is available.

---

debug info

CI timings

CI timings · View report · View job

Wall time 7m1s vs 5m54s (+18.9%). 5 job(s) slower, 8 faster, 1 unchanged.

• OS-specific tests / Windows-only tests: -34.0s
• Python lints / Windows footguns (blocking): +19.0s
• OS-specific tests / macOS-only tests: +6.0s
• Python lints / ruff enforcement (blocking): +5.0s
• Python tests / Run tests: -4.0s