Add manual E2E Jest test for workflow run lifecycle

[issuetopr-dev]

Summary

• Adds a manual end-to-end Jest test that exercises the full workflow run lifecycle for autoResolveIssue.
• The test enqueues a job via the Next.js API route, then polls the Workflow Runs list and events APIs to verify events are persisted and the workflow completes.

What’s included

• tests/config/jest.config.workflow.e2e.ts: Dedicated Jest config for E2E workflow tests (20 min timeout).
• tests/e2e/workflow-run.workflow.e2e.test.ts: The E2E test itself.
• package.json: Adds script pnpm run test:e2e:workflow.

How it works

1. Mocks next-auth’s auth() to provide a minimal session profile (GitHub login) needed by the enqueue route.
2. Calls POST /api/queues/[queueId]/jobs with name=autoResolveIssue and your repo/issue.
3. Polls GET /api/workflow-runs?repo=&issue= to discover the latest run for the issue.
4. Polls GET /api/workflow-runs/[workflowId]/events until it sees workflow.completed.
5. Asserts event persistence and completion. (Optionally logs status messages.)

Environment and execution

• This test is intended to be run manually and is skipped by default.
• To run, ensure your local stack is running (like pnpm dev): Redis, Neo4j, workers, Next app, and that GitHub/OpenAI credentials are set.

Required env vars for the test:

• RUN_WORKFLOW_E2E=true
• E2E_REPO_FULL_NAME=owner/repo (e.g., youngchingjui/openai-realtime-agents-test-playground)
• E2E_ISSUE_NUMBER=123
• E2E_GITHUB_LOGIN=your-github-username (used to mock auth())

Optional:

• E2E_BRANCH=feature/e2e-test-branch

Run it:

• pnpm run test:e2e:workflow

Notes

• We intentionally avoid hard-asserting PR creation to reduce flakiness across environments, but the workflow is designed to create a PR and you can verify on GitHub after the run completes.
• The test only runs when RUN_WORKFLOW_E2E=true to prevent accidental execution in CI (costly and slow).

Closes #1460

---

Updates in response to review

• Remove RUN_WORKFLOW_E2E gate: The E2E test now runs exclusively under its own Jest project via pnpm test:e2e:workflow, so no special env flag is required. This keeps it out of CI by default.
• Preflight checks: Added lightweight checks at test start to verify:
  • Required env vars for the run (E2E_REPO_FULL_NAME, E2E_ISSUE_NUMBER)
  • Redis availability (REDIS_URL + PING)
  • Neo4j availability (simple health check)
    If a check fails, the test exits early with a clear console warning and setup reminders.
• Test env example: Added __tests__/env.workflow.e2e.example (non-sensitive) to document/test variables. The test will auto-load __tests__/.env.workflow.e2e if present.
• Clarify Next.js origin: For the workflow-runs list call, use http://localhost:3000 and add a comment explaining that the origin is only used to construct a URL for request.nextUrl (no HTTP request is made).

Notes

• We kept the PR-creation assertion optional to avoid flakiness. If desired, we can add a follow-up flag to stub PR creation at the tool-call boundary for fully local runs.

[coderabbitai]

Important

Review skipped

Bot user detected.

To trigger a single review, invoke the @coderabbitai review command.

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[youngchingjui]

I wonder if we could mock the PR creation write at the tool call boundary. For example, I expect the LLM agent to call the create PR tool call. And when that's called, we don't need to actually make a PR - we can mock the actual execution of that tool.

[youngchingjui]

Where are these env variables saved? If they're in the tests folder (not root), then that's good. We should create a .env.example file if it doesn't already exist with default non-sensitive values.

[youngchingjui]

Instead of using an env variable to check if "shouldRun", I was expecting this to be another class of tests, where we would run something like "pnpm test:e2e" just like we do with the neo4j tests. So then we wouldn't need this env variable.

Seems like there's a "--selectProjects" option that might be useful.

[youngchingjui]

Does this work without specifying the port for NextJS? Usually 3000. Double check.

[youngchingjui]

Is there a way to check if these services are running and exit early if they're not? With clear messaging and reminders saying that these services need to be running

[youngchingjui]

Take this var out. Just use pnpm test:e2e:workflow and wrap these e2e workflows under a project or something so they're not normally run during CI