AgentSkillsCN

tdd

采用变异测试与覆盖率追踪，践行测试驱动开发流程

SKILL.md
--- frontmatter
name: tdd
description: Execute a test-driven development workflow with mutation testing and coverage tracking
user-invocable: true
allowed-tools: Read, Edit, Write, Grep, Glob, Bash
argument-hint: "[feature-name]"
hooks:
  - type: command
    command: |
      node -e "
        const fs = require('fs');
        const found = ['package.json','pytest.ini','pyproject.toml','Cargo.toml','go.mod'].filter(f => { try { fs.accessSync(f); return true; } catch { return false; } });
        if (found.length === 0) { console.log(JSON.stringify({additionalContext:'WARNING: No test framework config found. Install a test framework before starting TDD.'})); }
        else if (found.includes('package.json')) {
          const pkg = JSON.parse(fs.readFileSync('package.json','utf8'));
          const deps = {...pkg.devDependencies,...pkg.dependencies};
          const fw = ['vitest','jest','mocha','ava'].find(t => deps && deps[t]);
          if (!fw) console.log(JSON.stringify({additionalContext:'WARNING: No JS test framework in dependencies. Run npm install -D vitest (or jest) first.'}));
        }
      "
    once: true

Test-Driven Development Workflow

Implement $ARGUMENTS using strict TDD methodology.

Dynamic Context

  • •Test framework: !ls package.json 2>/dev/null && node -e "const p=require('./package.json');const d={...p.devDependencies,...p.dependencies};const f=['vitest','jest','mocha','ava','playwright','cypress'].find(t=>d[t]);console.log(f||'unknown')" 2>/dev/null || ls pytest.ini setup.cfg pyproject.toml Cargo.toml go.mod 2>/dev/null | head -1 || echo "unknown"
  • •Existing test patterns: !find . -name '*.test.*' -o -name '*.spec.*' -o -name '*_test.*' 2>/dev/null | head -5 || echo "No tests found"
  • •Source structure: !ls src/ app/ lib/ 2>/dev/null | head -10 || echo "N/A"

Red-Green-Refactor Protocol

Phase 1: RED - Write Failing Test

  1. •Create test file following project naming convention
  2. •Write one test case for the simplest expected behavior
  3. •Run test suite — confirm the test fails with expected error
  4. •If test passes without implementation, the test is wrong — rewrite it

Phase 2: GREEN - Minimal Implementation

  1. •Write the minimum code to make the failing test pass
  2. •No optimization, no generalization, no cleanup
  3. •Run test suite — confirm all tests pass (new + existing)
  4. •If any test fails, fix implementation (not the test)

Phase 3: REFACTOR - Improve Design

  1. •Remove duplication in both test and production code
  2. •Improve naming, extract methods, simplify logic
  3. •Run test suite after every change — maintain green
  4. •Apply SOLID principles where applicable

Phase 4: REPEAT

Return to Phase 1 with the next behavior. Continue until feature is complete.

Cycle Discipline

  • •One assertion per test (except closely related state checks)
  • •No production code without a failing test first
  • •No refactoring while red — get green first, then refactor
  • •Commit after each green phase (small, reversible increments)

Completion Criteria

Before declaring done:

  • • All acceptance criteria covered by tests
  • • Edge cases tested (null, empty, boundary, invalid input)
  • • Error paths tested (exceptions, failure modes)
  • • No skipped or commented-out tests
  • • All tests pass in isolation and as suite
  • • Existing tests unbroken (no regressions)

Per-Cycle Report

After each Red-Green-Refactor cycle, report:

code
Cycle [N]:
  Test: [test name and file]
  Status: RED → GREEN → REFACTORED
  Implementation: [file:line changed]
  Tests passing: [X/Y]

Final Summary

code
Feature: $ARGUMENTS
Cycles completed: [N]
Tests added: [N]
Files created: [list]
Files modified: [list]
All tests passing: YES/NO