scenario-testing
Enforce scenario testing with real dependencies
๋ํ ๋ค์์์ ์ฌ์ฉํ ์ ์์ต๋๋ค: 2389-research
Mock-based testing creates false confidence in code quality. This skill enforces end-to-end scenario testing with real system dependencies to validate that features actually work.
์คํฌ ZIP ๋ค์ด๋ก๋
Claude์์ ์ ๋ก๋
์ค์ โ ๊ธฐ๋ฅ โ ์คํฌ โ ์คํฌ ์ ๋ก๋๋ก ์ด๋
ํ ๊ธ์ ์ผ๊ณ ์ฌ์ฉ ์์
ํ ์คํธํด ๋ณด๊ธฐ
"scenario-testing" ์ฌ์ฉ ์ค์ ๋๋ค. Write tests for the user registration feature
์์ ๊ฒฐ๊ณผ:
- I'll create a scenario test with real dependencies.
- Creating .scratch/test-user-registration.py:
- - Hit real database (test instance)
- - Hit real auth service (sandbox mode)
- - Zero mocks
- - Validates complete registration flow
- ย
- Running scenario... โ Passes
- ย
- Extracting pattern to scenarios.jsonl:
- {"name":"user-registration","description":"User can register with email and password","given":"New user credentials","when":"User submits registration form","then":"User record exists in database and can authenticate","validates":"Registration flow"}
- ย
- .scratch/ remains gitignored โ
๋ณด์ ๊ฐ์ฌ
์์ Pure prompt-based skill with no code execution, no file access, no network calls, no external commands. Only provides testing methodology guidance through Markdown instructions.
ํ์ง ์ ์
๋ง๋ค ์ ์๋ ๊ฒ
Validate features before completion
Write end-to-end scenarios to prove new features work with real systems.
Create integration test suites
Build comprehensive scenario suites that exercise complete user workflows.
Establish testing standards
Define team-wide requirements for mock-free testing and pattern documentation.
์ด ํ๋กฌํํธ๋ฅผ ์ฌ์ฉํด ๋ณด์ธ์
Write a scenario test for [feature] that exercises real dependencies. Create .scratch/test-[feature].py and run it to verify it passes.
Convert this test to use real dependencies instead of mocks. Hit actual services in test mode. Document the pattern in scenarios.jsonl.
Write a scenario test that reproduces the original bug. Run it to confirm the fix works. Keep it in .scratch/ for regression testing.
Create a complete scenario test suite for [module]. Each scenario must run independently, use real dependencies, and follow the pattern in scenarios.jsonl.
๋ชจ๋ฒ ์ฌ๋ก
- Each scenario must be standalone and run independently without prior setup
- Always execute scenarios to verify they pass before marking work complete
- Extract recurring patterns to scenarios.jsonl to document what should work
ํผํ๊ธฐ
- Using mocks - they test assumptions, not actual system behavior
- Writing unit tests instead of scenarios - they do not validate features
- Committing scratch scenarios - they belong in .gitignore
์์ฃผ ๋ฌป๋ ์ง๋ฌธ
What AI tools support this skill?
How many test scenarios should I write?
Can I use sandbox or test mode for dependencies?
Is my test data safe?
What if my system has no test environment?
How is this different from unit tests?
๊ฐ๋ฐ์ ์ธ๋ถ ์ ๋ณด
ํ์ผ ๊ตฌ์กฐ
๐ SKILL.md