scenarios_skill

Overview

This skill validates features end-to-end against real dependencies with no mocks. It focuses on scenario testing: self-contained proof programs, run against actual databases, APIs, and services, and documented as reusable patterns.

How this skill works

Write a disposable script in a local .scratch/ directory that sets up real state, executes real requests or operations, asserts on actual responses, and performs cleanup. Successful patterns are recorded as one-line JSON objects in scenarios.jsonl so teams can reuse proven integrations and detect regressions.

When to use it

• Validating full user flows end-to-end without simulation
• Testing integrations across services and external APIs
• Verifying authentication, API contracts, or real workflows
• Proving or reproducing bugs with real dependencies
• Documenting known-good behavior for regression detection

Best practices

• Always run scenarios against real implementations (local, staging, or test accounts) — NO MOCKS EVER
• Keep proof programs self-contained and runnable with a single command
• Ensure .scratch/ is gitignored before creating disposable scripts
• Include setup and thorough cleanup stages to avoid leftover test data
• Document successful runs in scenarios.jsonl with clear setup, steps, expected result, and tags

Example use cases

• Run a login-to-profile flow with a real database and assert JWT behavior
• Validate third-party webhook processing by exercising a live API endpoint
• Reproduce a production bug locally using real service instances to gather evidence
• Confirm an end-to-end payment flow against a sandboxed payment provider
• Capture a flaky integration as a documented scenario to prevent regressions