[sylvesterkaczmarek]

Summary

Fixes #781. captureTurn() currently buffers every notification while state.turnId is falsy. If turn/start returns without turn.id, the buffer is flushed once, but all later notifications, including turn/completed, are buffered again forever. The task remains stuck running even though the server completed it.

• Track whether the start request has responded independently of turn.id.
• Buffer only until the start response is processed; then forward subsequent notifications through the normal thread/turn routing.
• Preserve the existing pre-response buffer flush, previous notification handler forwarding, and cleanup. completeTurn() still adopts the ID from the completion notification if it was omitted from the response.
• Extend the fake app-server to reproduce two scenarios: a missing turn.id with later notifications, and a pre-response turn/started followed by post-response completion.
• Add bounded end-to-end tests asserting the task finishes successfully rather than hanging indefinitely.

Verification

• 93 tests passed (npm test), including 2 new real CLI/fake-server integration tests for the missing-turn-ID cases.
• Focused Node syntax checks on the three changed files and Git whitespace checks passed.
• No production app-server or account credentials were used.

This is independent of the existing PRs for terminal error handling and inferred completion, which address different failure triggers.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-09T15:23:45.754864Z	80fe237	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.