How can I automate smoke tests for my API after deploys?

A good way to automate API smoke tests after deploys is to keep them small, fast, and deployment-gated: after a new build is deployed, run a handful of “does the API basically work?” checks against the deployed environment and fail the pipeline if any fail.

1) Decide what a smoke test should cover

Smoke tests should verify only the most critical paths, for example:

• Service is up: GET /health or GET /status
• Auth works: obtain token or use a test token
• One or two core endpoints respond correctly
• A minimal write/read flow if your API is stateful
• Basic dependency checks if relevant (DB-backed endpoint, cache, etc.)

Keep them short enough to run in seconds, not minutes.

---

2) Run them against the deployed environment

The key is to run tests after deployment and point them to the actual deployed base URL.

Typical environments:

• staging
• ephemeral preview environment
• production canary or post-deploy verification endpoint

Use environment variables like:

• API_BASE_URL=https://staging.example.com
• API_TOKEN=...

---

3) Implement them in a simple test runner

Common choices:

• Postman/Newman
• pytest + requests
• JUnit/TestNG
• Playwright/API testing
• k6 for lightweight HTTP checks
• Bash + curl for very simple setups

Example with pytest

import os
import requests

BASE_URL = os.environ["API_BASE_URL"]
TOKEN = os.environ["API_TOKEN"]

def auth_headers():
    return {"Authorization": f"Bearer {TOKEN}"}

def test_health():
    r = requests.get(f"{BASE_URL}/health", timeout=5)
    assert r.status_code == 200
    assert r.json()["status"] == "ok"

def test_list_items():
    r = requests.get(f"{BASE_URL}/items", headers=auth_headers(), timeout=5)
    assert r.status_code == 200
    assert isinstance(r.json(), list)

---

4) Wire it into your CI/CD pipeline

After the deploy step, add a smoke-test job.

Example GitHub Actions

name: Deploy and Smoke Test

on:
  push:
    branches: [main]

jobs:
  deploy:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - name: Deploy
        run: ./deploy.sh

  smoke-test:
    runs-on: ubuntu-latest
    needs: deploy
    steps:
      - uses: actions/checkout@v4
      - name: Install deps
        run: pip install -r requirements.txt
      - name: Run smoke tests
        env:
          API_BASE_URL: https://staging.example.com
          API_TOKEN: ${{ secrets.API_TOKEN }}
        run: pytest tests/smoke -q

If deploy and smoke test are in the same workflow, the pipeline can fail fast and alert you immediately.

---

5) Use deployment-specific gate behavior

Depending on your deployment strategy:

• Blue/green: run smoke tests on the new environment before switching traffic
• Canary: deploy to a small slice, then run smoke tests before full rollout
• Rolling: run smoke tests once rollout completes
• Preview environments: run smoke tests on every PR deployment

This is especially useful to catch config or migration issues before users see them.

---

6) Make tests reliable

Smoke tests should be stable and low-noise:

• Use test data that won’t collide with real users
• Prefer idempotent operations
• Add timeouts and retries only where appropriate
• Avoid brittle assertions on dynamic values
• Clean up any resources you create
• Don’t depend on external services unless necessary

---

7) Report failures clearly

If a smoke test fails, you want immediate, actionable feedback:

• Which endpoint failed
• HTTP status code
• Response body snippet
• Correlation/request ID if available
• Which deployment version was tested

This helps distinguish app issues from deployment issues quickly.

---

8) Optional: add post-deploy monitoring checks too

Smoke tests are great, but combine them with:

• synthetic monitoring
• uptime checks
• alerting on error rates/latency
• log/trace inspection after rollout

Smoke tests verify “it works now”; monitoring tells you “it keeps working.”

---

Recommended minimal setup

If you want the simplest practical solution:

1. Add a /health endpoint
2. Create 3–5 API smoke tests with pytest or Newman
3. Store base URL and tokens as CI secrets/env vars
4. Run the suite automatically after deploy
5. Fail the deployment pipeline if tests fail

If you want, I can show you:

• a GitHub Actions example,
• a GitLab CI example,
• a Postman/Newman setup,
• or a pytest-based smoke test template for your API.