[hellices]

Closes #478

Summary

Two operation journeys failed on main's test (3.11) (run 39980913476) with NoMatches: No nodes match '.confirm-title' on ConfirmScreen() from the eval driver's handle(). The second NoMatches in that log, 'AgentPanel' on Screen(id='_default'), came from the journey tearing down after the first. #477 fixes that teardown error on its own.

The driver's readiness check returns as soon as app.screen is a ConfirmScreen, and handle() read the dialog's title straight away. Textual makes a pushed screen current before it composes the screen's widgets, so on a loaded runner the poll can land in between.

• The fix: handle() now first waits, through _await_dialog_composed, until the current dialog has its title or is no longer current.
• A dialog that closes first falls through to the existing closed-before-a-decision path, so expiry handling (Eval test_an_expired_approval_is_terminal_and_mutates_nothing flakes on loaded runners #451, test: note each approval dialog when Textual pushes it #460) is unchanged.

Verification

tests/evals/test_operation_journeys.py::test_a_dialog_still_composing_is_read_once_it_has_its_widgets delays ConfirmScreen's compose by 0.3 seconds. That holds the dialog in the pushed-but-empty state the runner hit. The test expects one matched dialog_observed and no hard failures.

• RED: without the wait it failed three runs out of three with CI's exact pair of errors.
• GREEN: with the wait it passes three runs out of three.

Run locally on macOS with Python 3.13 and the frozen lock. Each step ran directly with the project's environment rather than through make, because this machine cannot sync from PyPI:

Command	Result
pytest -p no:tach tests/evals/	556 passed
python scripts/check_source_size.py	passed
ruff check src/ tests/ and ruff format --check src/ tests/	passed; 659 files already formatted
SKIP=mypy,source-size pre-commit run --files <changed files>	passed
mypy	no issues in 659 source files
tach check	all modules validated
pytest -n 8 -q	13,750 passed, 25 skipped

Checklist

• This closes an accepted issue in this repository that is assigned to me.
• Tests cover the change, written to fail first where behavior changed.
• make check passes locally.
• User-facing docs and the README match the change, or nothing user-facing changed.
• No security invariant is weakened: writes still need a keystroke approval
  and a fail-closed audit entry, and sensitive reads are still masked.

[Copilot]

🟢 Approval recommended

The focused test-only fix addresses the composition race, preserves expiry handling, and has targeted regression coverage with no identified blocking issues.

No open issues 🎉

What changed in this PR

This test-only change addresses #478 by preventing the operation-journey driver from reading an approval dialog before Textual has composed its widgets.

Changes:

• Wait for the dialog’s title to exist or for the dialog to close, preserving existing closed-before-decision handling.
• Add a regression test that delays composition and verifies successful dialog observation without hard failures.

File	Description
tests/​evals/​test_operation_journeys.py	Adds regression coverage for delayed dialog composition.
tests/​evals/​operation_app.py	Waits for dialog composition before reading its contents.

_{Review effort: Balanced}

[my-reviewer-agent]

검토 범위에서 보고할 이슈가 없습니다.

[hellices]

Superseded by #482, which carries this pull request's commits unchanged (b72b03a) together with the other open fixes and the dependency security bumps, so all of them need one CI run and one merge instead of a rebase after each merge. Closing in favour of #482; the branch stays, so this can be reopened if needed.