fix(server): block V2 replacements until session cleanup completes

[shivamhwp]

V2 marks a provider session stopped after a 30-second cleanup timeout even when the provider is still shutting down. That allows a replacement to start too early and can revoke credentials still held by the old process.

Keep ownership of the pending cleanup until it finishes. A timeout records an error and blocks replacements and new attachments. Close, detach, and provider-wide close retries wait on that same cleanup. Shared credentials survive until the last holder finishes, and an attachment racing with close fails without revoking a peer's credential. Successful cleanup clears the error and block; failed cleanup keeps replacements blocked until the environment restarts.

Targets the Orchestration V2 branch in #2829. Includes main through b70015b6db and the V2 target through a8cc38b95b. The main integrations are separate merge commits; the original cleanup commits remain isolated in the history. The aggregate PR diff now includes main's web, mobile, server, and shared-package updates.

The merge keeps V2's run and projected-item model while carrying forward main's pasted-text attachments, connection controls, notifications, and response streaming settings. Paragraph streaming and missing-worktree fallback are adapted to the V2 execution and launch paths; retired V1 runtime files stay removed. Main advanced to 73b206f4bf during validation; that later worktree-setup feature is not included. The latest integration also includes main's CLI release archives and update/uninstall commands, and its compact-sidebar revert.

Validation on 8023b0a1d8: focused server, web, mobile, and client-runtime tests pass, including session cleanup, V2 replay, paragraph boundaries and completion, and worktree fallback. Typechecks pass for server, web, desktop, mobile, contracts, client-runtime, and shared. Targeted lint, formatting, and the exact unused-code check pass. Follow-up dfdde1f624 passed all GitHub CI jobs after fixing Hermes array compatibility and migrating notification test fixtures to V2. Latest head 96704fdb6b additionally passes 207 focused tests, server/web/desktop typechecks, CLI bundle and help smoke checks, marketing build, targeted lint, formatting, and unused-code checks. All GitHub CI jobs pass on this latest head. Vercel still reports a deployment failure without a log URL, so the PR is not fully green. Macroscope's last correctness review was skipped because its $20.01 estimate exceeded the $15 per-review limit; human review remains needed. No new browser or device verification was performed for this integration; the evidence below predates the merge.

Earlier UI verification

Before: no warning while cleanup is still running.

After: the PR shows the cleanup timeout and blocks replacement sessions.

Recovery: the warning clears automatically after cleanup succeeds.

Captured in a real Chromium client connected to an isolated local server and the real Codex provider. Both the V2 base and PR head used a temporary five-second idle timer and a 120-second cleanup delay. The production 30-second cleanup timeout was unchanged.

• Before, V2 base 3ed8c4df: the session was marked stopped after the timeout, with no warning while cleanup was still running.
• After, PR head b0ccd56f: the session stayed in error and the app showed that replacement sessions were blocked until cleanup completed.
• Recovery: cleanup finished, the stored error cleared, and the warning disappeared without dismissal or a page reload.

Temporary test hooks were removed after capture and are not included in this PR. Original PNGs and the captured session-state evidence are retained with the test artifacts.

Earlier detailed test report. The validation above records the integrated commits.

Implemented and tested by GPT-6 using Codex in T3 Code for shivamhwp.

[shivamhwp]

Note: GPT-6 on behalf of shivam (@shivamhwp).

The first CI run found failures in unchanged V2-base files: Check reports formatting in seven files, Test Server 1 expects 58 migrations but the base registers 59, and Release Smoke fails on an unused expo-notifications@57.0.15 patch. None of those files are changed by this PR. Keeping those fixes separate from session cleanup.

The review fixes are in d68d735a. All 34 focused session-manager and turn-start tests pass locally, along with targeted lint and server typecheck. Waiting for the new CI and review results.

[cursor]

Stale Bugbot comment from a previous run.

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — This PR is a broad product and platform change rather than a contained session-cleanup fix: it adds Forgejo/GitHub routing, reusable authentication, mobile widgets and paste workflows, notifications, and substantial orchestration changes. It also changes product defaults and authentication behavior, so the scope and sensitivity warrant human review.

Not approved because:

• Per-review cost limit exceeded (workspace setting). Approvability relies on correctness review in order to determine eligibility

^{Review your spending limits in Billing settings, or comment @macroscope-app review this PR to bypass the limit and review now. You can add or adjust custom eligibility rules. Learn more.}

[shivamhwp]

Note: GPT-6 on behalf of shivam (@shivamhwp).

Latest head bf93d875 passes Bugbot and Macroscope correctness. All reported review threads are resolved. Locally, 58 focused tests across six session-manager consumer test files pass, as do targeted lint and server typecheck.

CI remains blocked outside this change: seven unchanged files fail formatting, the migration test expects 58 entries instead of the registered 59, and Release Smoke rejects the unused Expo notifications patch. Earlier Test Server 3 logs also show stale replay expectations for the runtime-info prompt suffix. I reproduced the migration failure, the seven formatting failures, and the tool_call_read_only/cursor replay mismatch locally. The relevant adapter, migration, fixture, and release files match the V2 base.

Leaving these base fixes out of the cleanup PR pending maintainer direction. Nothing has been merged.

[cursor]

Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit bc10499e4e572b4e7f364ee5d8077acdb6f6ad92. Configure here.}

[juliusmarminge]

Stale duplicate of #11492 — the same session-cleanup ownership fix exists on the newer branch. Thanks!