API Test Plan Skill

APIs fail in specific, testable ways: wrong status codes, schema drift, missing auth checks, sloppy validation, unhelpful errors. This skill plans the tests that catch them — per endpoint, across the response codes and the error paths, with contract checks so the API keeps its promises to clients. It tests the whole behaviour, not just the happy 200.

Working from a brief

Given an endpoint or an API description, produce the test plan anyway — infer the likely parameters, responses, auth model, and error cases, labelling assumptions. Always include auth, validation, and negative cases. Never hand back a question instead of a plan.

Required Inputs

Ask for these only if they aren't already provided (else infer and label):

• The API — REST/GraphQL, the endpoints/operations, and what they do.
• Contract — request/response schemas, parameters, status codes (or an OpenAPI/spec if available).
• Auth & rules — the auth model (token/scopes/roles), rate limits, and validation rules.
• Dependencies & data — downstream services, and the data/state needed to test.

Output Format

API Test Plan: [API / endpoint]

Per endpoint, a set of cases grouped by type:

ID	Endpoint	Case	Type	Request	Expected status	Expected body / assertion
API-01	POST /orders	valid create	Functional