[waleedlatif1]

Summary

Adds a live acceptance case to saved-run-read.live.test.ts: a long draft test run goes through the real controller against the worker's sliced synchronous workflows run. It builds on #9464 (workflows run --wait-timeout and the workflow-run-wait-timeout capability) and #9469 (a still-running slice settles as done).

The draft is Start → Wait 300 s → Wait 200 s → Function returning {invoice: 42, settled: "after-waits"}. It runs through runCopilotLifecycle, the sim_cli handler, the embedded CLI, and the v2 execute and runs routes, against the worker's scripted probe (tools/probes/src/draft-run-slices.ts).

The case has three variants:

• declared: the run detaches after 240 s with exit 4, continues as two workflows runs wait slices, and is read back with workflows runs get --include-output. No call holds the run (each under 250 s), no slice row is failed, and the model receives the run's real output.
• undeclared: one held workflows run --manual with no --wait-timeout, exit 0, with the output.
• own-budget: the model's own --wait-timeout 5 reaches it as exit 4, with the running receipt and the "Keep waiting" hint.

Two fixture changes make the controller cases runnable on a self-hosted default:

• The harness pins direct Sim transport. The control fixture answers run control on its direct endpoint; under the checkpoint default, run control went unanswered and the turn timed out.
• The fixture plan's execution timeout is set per case, so a run can outlast a slice.

Opt-in like the other controller cases: SIM_HELPERS_SMOKE=1, MSHIP_TEST_DATABASE_URL, MSHIP_WORKER_ROOT. Optionally MSHIP_PROBE_REDIS_URL, and MSHIP_SLICES_REPORT_PATH for the JSON evidence. It takes about nine minutes.

Companion: simstudioai/mothership#872

Type of Change

• Bug fix
• New feature
• Breaking change
• Documentation
• Other: live test

Testing

Ran at this head against the worker branch feat/slice-draft-test-runs (70d213247), 09:55–10:04 PT. All three variants pass:

• declared: 240.1 s (exit 4) → 240.0 s (exit 4) → 29.3 s (exit 0) → read-back 0.04 s; turn 509.9 s.
• undeclared: one call, 260.4 s, exit 0.
• own-budget: one call, 5.1 s, exit 4.

The case only adds a test and fixture knobs; no production code changes.

Checklist

• Code follows project style guidelines
• Self-reviewed my changes
• Tests added/updated and passing
• No new warnings introduced
• I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
docs	Ready	Preview	Oct 30, 2026 5:21pm UTC

[github-actions]

⚠️ Cross-repo companion check

One or more companion PRs aren't merged into staging yet. Merging this without them will leave copilot and sim out of sync — merge them in lockstep.

• ❌ simstudioai/mothership#872 — OPEN, not merged (targets staging) — feat(worker): wait on a synchronous workflows run in slices on a Sim that can de

[cubic-dev-ai]

No issues found across 1 file

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.

_{Turn on auto-fix | Re-trigger cubic}

[greptile-apps]

Confidence Score: 4/5

[Low impact] The changes affect an opt-in live test, not production behavior.

Satisfy the repository's shared-helper and non-null assertion requirements before merging; the startup deadline is a non-blocking improvement.

Findings

1.  Delay bypasses the shared helper ▶
2.  Non-null assertion bypasses required checks ▶
3.  Worker startup can stall ▶

Summary

Adds an opt-in live test that runs a draft workflow through the real controller.

• The live test follows long draft runs through the real controller.

Diagram

%%{init: {'theme': 'neutral'}}%%
flowchart TD
  A[Real controller starts draft run] --> B{Capability declared?}
  B -->|No| C[Hold one call until completion]
  B -->|Yes| D{Model supplied wait budget?}
  D -->|Yes| E[Return running receipt after five seconds]
  D -->|No| F[Return running receipt after 240 seconds]
  F --> G[Wait in further slices]
  G --> H[Read completed run with output]
  C --> I[Check model received output]
  H --> I

Loading

_{Reviews (1) · Last reviewed commit: "test(mothership): drive a draft run past..." · Reviewed by Greptile}