sceptre-mcp-server: Instructions file for Kiro

TDD Workflow Skill

Activate this skill when implementing new features, fixing bugs, or refactoring Python code.

The Cycle: RED → GREEN → REFACTOR

Step 1: Define the Interface (SCAFFOLD)

• Define type hints and function signatures
• Create function/class stubs that raise NotImplementedError
• Clarify the contract before writing any logic

Step 2: Write Failing Tests (RED)

• Write tests that describe the desired behavior
• Cover happy path, edge cases, and error scenarios
• Run tests — they MUST fail (poetry run pytest -q)

Step 3: Implement Minimal Code (GREEN)

• Write the simplest code that makes all tests pass
• Don't optimize yet — just make it work
• Run tests — they should all pass now

Step 4: Refactor (IMPROVE)

• Clean up the implementation while keeping tests green
• Extract constants, improve naming, reduce duplication
• Run tests after each refactoring step

Step 5: Verify Coverage

• Check coverage is at 80%+: poetry run pytest --cov --cov-report=term-missing -q
• Add tests for any uncovered paths
• Focus on meaningful coverage, not just line count

Test Structure Template

import pytest
from unittest.mock import MagicMock, patch


class TestModuleName:
    """Tests for [Module/Function Name]."""

    def test_happy_path_when_valid_input(self):
        # Arrange
        input_data = {...}
        # Act
        result = function_under_test(input_data)
        # Assert
        assert result == expected

    def test_raises_value_error_on_empty_input(self):
        with pytest.raises(ValueError, match="Input cannot be empty"):
            function_under_test({})

    @pytest.mark.parametrize("input,expected", [
        (None, None),
        ("", ""),
        ("valid", "VALID"),
    ])
    def test_edge_cases(self, input, expected):
        assert function_under_test(input) == expected

Rules

• NEVER write implementation before tests
• NEVER modify tests to make them pass (fix the implementation)
• Each test should test ONE behavior
• Tests should be independent (no shared mutable state between tests)
• Mock external dependencies (boto3, opensearch, slack_sdk), not internal logic