API endpoint tester

A tiny smoke test runner with no dependencies. It sends real HTTP requests, so only point it at services the user owns or has permission to test.

When to use

• After a deploy, to confirm the main endpoints respond.
• When the user suspects an endpoint changed shape or status code.
• To turn a list of curl commands into a repeatable check.

How to run

python scripts/check_api.py checks.json --base-url https://staging.example.com

The checks file is a JSON list:

[
  {"name": "health", "method": "GET", "path": "/health", "expect_status": 200, "expect_body_contains": "ok"},
  {"name": "create", "method": "POST", "path": "/items", "json": {"name": "demo"}, "expect_status": 201}
]

Optional fields: headers (object), json (request body), expect_status (default 200), expect_body_contains (string) and timeout (seconds, default 10).

Output

JSON with base_url, passed, failed and results. Each result has name, ok, status, elapsed_ms and, when it failed, a reason. The exit code is 1 when any check fails.

After running

Report failures first with the reason and status code. Mention slow endpoints (over one second). Do not retry failing POST requests automatically, they may create duplicates.

Errors

A missing checks file exits with code 2. Invalid JSON or a malformed check exits with code 3. Connection errors are reported as failed checks, not crashes.