---
title: "Deployment checks"
description: "How CI verifies the Vercel build artifact, and how to run the deployment smoke test against Vercel or Docker."
---

The **Vercel compatibility** workflow (`.github/workflows/vercel.yml`) has two jobs.

## Artifact build

Runs on every pull request and every push to `main`. It needs no Vercel account and no secrets, so fork pull requests run it safely.

The job:

1. Starts an isolated PostgreSQL service.
2. Writes a local `.vercel/project.json` with the `services` framework and runs `vercel build --prod` offline, with placeholder Blob credentials. This also applies migrations to the isolated database.
3. Checks the `backend` service Function:
   - runtime is `nodejs24.x`, `maxDuration` is `300`, and the handler is `apps/server/vercel.mjs`;
   - a copy of the Function outside the checkout loads with `--no-experimental-require-module`, which matches the Vercel runtime, so a dependency the build left out fails the job;
   - the copy rejects an unauthenticated staging request and serves the prerendered homepage.

If a new server dependency fails the loading check, add it and its own dependencies to `bundledInteropPackages` in `apps/server/tsdown.config.ts`. CommonJS dependencies need this: Vercel's service builder loads them through pnpm links that it leaves out of the Function.

## Live smoke test

Runs only when started manually (`workflow_dispatch`). It uses the `vercel-smoke` GitHub environment and runs `tooling/deployment/smoke.mjs` against `VERCEL_SMOKE_URL`.

<Warning>
  Point the smoke test only at a dedicated test installation. It creates an account, a public resume, and files, then deletes them.
</Warning>

The script checks health, public pages, signup, resume CRUD, public PDF rendering, a 10 MiB upload and download, and one-time use of staged requests.

To also check a 25 MiB agent attachment, configure a deterministic OpenAI-compatible test provider that serves the model `smoke-model`:

| Name | Kind | Value |
| --- | --- | --- |
| `VERCEL_SMOKE_URL` | Variable | Test installation origin |
| `VERCEL_SMOKE_AI_BASE_URL` | Variable | Test provider base URL |
| `VERCEL_SMOKE_AI_API_KEY` | Secret | Test provider API key |

No paid AI model is needed.

## Run the smoke test locally

Against a local Docker installation:

```bash
SMOKE_URL=http://localhost:3000 node tooling/deployment/smoke.mjs
```

Add `SMOKE_AI_BASE_URL` and `SMOKE_AI_API_KEY` to include the attachment check.