AGENT SKILL

tdd-guide

Test-driven development skill for writing unit tests, generating test fixtures and mocks, analyzing coverage gaps, and guiding red-green-refactor workflows across Jest, Pytest, JUnit, Vitest, and Mocha. Use when the user asks to write tests, improve test coverage, practice TDD, generate mocks or stubs, or mentions testing frameworks like Jest, pytest, or JUnit.

License 许可
MIT
最近更新
2026-08-16
Stars
★ 0
Forks
0

01 前言 Front Matter

name
tdd-guide
description
Test-driven development skill for writing unit tests, generating test fixtures and mocks, analyzing coverage gaps, and guiding red-green-refactor workflows across Jest, Pytest, JUnit, Vitest, and Mocha. Use when the user asks to write tests, improve test coverage, practice TDD, generate mocks or stubs, or mentions testing frameworks like Jest, pytest, or JUnit.

02 资源文件 17

03 SKILL.md 正文

TDD Guide

Test-driven development skill for generating tests, analyzing coverage, and guiding red-green-refactor workflows across Jest, Pytest, JUnit, and Vitest.


Workflows

Generate Tests from Code

  1. Provide source code (TypeScript, JavaScript, Python, Java)
  2. Specify target framework (Jest, Pytest, JUnit, Vitest)
  3. Run test_generator.py with requirements
  4. Review generated test stubs
  5. Validation: Tests compile and cover happy path, error cases, edge cases

Analyze Coverage Gaps

  1. Generate coverage report from test runner (npm test -- --coverage)
  2. Run coverage_analyzer.py on LCOV/JSON/XML report
  3. Review prioritized gaps (P0/P1/P2)
  4. Generate missing tests for uncovered paths
  5. Validation: Coverage meets target threshold (typically 80%+)

TDD New Feature

  1. Write failing test first (RED)
  2. Run tdd_workflow.py --phase red to validate
  3. Implement minimal code to pass (GREEN)
  4. Run tdd_workflow.py --phase green to validate
  5. Refactor while keeping tests green (REFACTOR)
  6. Validation: All tests pass after each cycle

Examples

Test Generation — Input → Output (Pytest)

Input source function (math_utils.py):

def divide(a: float, b: float) -> float:
    if b == 0:
        raise ValueError("Cannot divide by zero")
    return a / b

Command:

python scripts/test_generator.py --input math_utils.py --framework pytest

Generated test output (test_math_utils.py):

import pytest
from math_utils import divide

class TestDivide:
    def test_divide_positive_numbers(self):
        assert divide(10, 2) == 5.0

    def test_divide_negative_numerator(self):
        assert divide(-10, 2) == -5.0

    def test_divide_float_result(self):
        assert divide(1, 3) == pytest.approx(0.333, rel=1e-3)

    def test_divide_by_zero_raises_value_error(self):
        with pytest.raises(ValueError, match="Cannot divide by zero"):
            divide(10, 0)

    def test_divide_zero_numerator(self):
        assert divide(0, 5) == 0.0

Coverage Analysis — Sample P0/P1/P2 Output

Command:

python scripts/coverage_analyzer.py --report lcov.info --threshold 80

Sample output:

Coverage Report — Overall: 63% (threshold: 80%)

P0 — Critical gaps (uncovered error paths):
  auth/login.py:42-58   handle_expired_token()       0% covered
  payments/process.py:91-110  handle_payment_failure()   0% covered

P1 — High-value gaps (core logic branches):
  users/service.py:77   update_profile() — else branch  0% covered
  orders/cart.py:134    apply_discount() — zero-qty guard  0% covered

P2 — Low-risk gaps (utility / helper functions):
  utils/formatting.py:12  format_currency()            0% covered

Recommended: Generate tests for P0 items first to reach 80% threshold.

Key Tools

Tool Purpose Usage
test_generator.py Generate test cases from code/requirements python scripts/test_generator.py --input source.py --framework pytest
coverage_analyzer.py Parse and analyze coverage reports python scripts/coverage_analyzer.py --report lcov.info --threshold 80
tdd_workflow.py Guide red-green-refactor cycles python scripts/tdd_workflow.py --phase red --test test_auth.py
fixture_generator.py Generate test data and mocks python scripts/fixture_generator.py --entity User --count 5

Additional scripts: framework_adapter.py (convert between frameworks), metrics_calculator.py (quality metrics), format_detector.py (detect language/framework), output_formatter.py (CLI/desktop/CI output).


Input Requirements

For Test Generation:

  • Source code (file path or pasted content)
  • Target framework (Jest, Pytest, JUnit, Vitest)
  • Coverage scope (unit, integration, edge cases)

For Coverage Analysis:

  • Coverage report file (LCOV, JSON, or XML format)
  • Optional: Source code for context
  • Optional: Target threshold percentage

For TDD Workflow:

  • Feature requirements or user story
  • Current phase (RED, GREEN, REFACTOR)
  • Test code and implementation status

Spec-First Workflow

TDD is most effective when driven by a written spec. The flow:

  1. Write or receive a spec — stored in specs/<feature>.md
  2. Extract acceptance criteria — each criterion becomes one or more test cases
  3. Write failing tests (RED) — one test per acceptance criterion
  4. Implement minimal code (GREEN) — satisfy each test in order
  5. Refactor — clean up while all tests stay green

Spec Directory Convention

project/
├── specs/
│   ├── user-auth.md          # Feature spec with acceptance criteria
│   ├── payment-processing.md
│   └── notification-system.md
├── tests/
│   ├── test_user_auth.py     # Tests derived from specs/user-auth.md
│   ├── test_payments.py
│   └── test_notifications.py
└── src/

Extracting Tests from Specs

Each acceptance criterion in a spec maps to at least one test:

Spec Criterion Test Case
"User can log in with valid credentials" test_login_valid_credentials_returns_token
"Invalid password returns 401" test_login_invalid_password_returns_401
"Account locks after 5 failed attempts" test_login_locks_after_five_failures

Tip: Number your acceptance criteria in the spec. Reference the number in the test docstring for traceability (# AC-3: Account locks after 5 failed attempts).

Cross-reference: See engineering/spec-driven-workflow for the full spec methodology, including spec templates and review checklists.


Red-Green-Refactor Examples Per Language

TypeScript / Jest

// test/cart.test.ts
describe("Cart", () => {
  describe("addItem", () => {
    it("should add a new item to an empty cart", () => {
      const cart = new Cart();
      cart.addItem({ id: "sku-1", name: "Widget", price: 9.99, qty: 1 });

      expect(cart.items).toHaveLength(1);
      expect(cart.items[0].id).toBe("sku-1");
    });

    it("should increment quantity when adding an existing item", () => {
      const cart = new Cart();
      cart.addItem({ id: "sku-1", name: "Widget", price: 9.99, qty: 1 });
      cart.addItem({ id: "sku-1", name: "Widget", price: 9.99, qty: 2 });

      expect(cart.items).toHaveLength(1);
      expect(cart.items[0].qty).toBe(3);
    });

    it("should throw when quantity is zero or negative", () => {
      const cart = new Cart();
      expect(() =>
        cart.addItem({ id: "sku-1", name: "Widget", price: 9.99, qty: 0 })
      ).toThrow("Quantity must be positive");
    });
  });
});

Python / Pytest (Advanced Patterns)

# tests/conftest.py — shared fixtures
import pytest
from app.db import create_engine, Session

@pytest.fixture(scope="session")
def db_engine():
    engine = create_engine("sqlite:///:memory:")
    yield engine
    engine.dispose()

@pytest.fixture
def db_session(db_engine):
    session = Session(bind=db_engine)
    yield session
    session.rollback()
    session.close()

# tests/test_pricing.py — parametrize for multiple cases
import pytest
from app.pricing import calculate_discount

@pytest.mark.parametrize("subtotal, expected_discount", [
    (50.0, 0.0),       # Below threshold — no discount
    (100.0, 5.0),      # 5% tier
    (250.0, 25.0),     # 10% tier
    (500.0, 75.0),     # 15% tier
])
def test_calculate_discount(subtotal, expected_discount):
    assert calculate_discount(subtotal) == pytest.approx(expected_discount)

Go — Table-Driven Tests

// cart_test.go
package cart

import "testing"

func TestApplyDiscount(t *testing.T) {
    tests := []struct {
        name     string
        subtotal float64
        want     float64
    }{
        {"no discount below threshold", 50.0, 0.0},
        {"5 percent tier", 100.0, 5.0},
        {"10 percent tier", 250.0, 25.0},
        {"15 percent tier", 500.0, 75.0},
        {"zero subtotal", 0.0, 0.0},
    }

    for _, tt := range tests {
        t.Run(tt.name, func(t *testing.T) {
            got := ApplyDiscount(tt.subtotal)
            if got != tt.want {
                t.Errorf("ApplyDiscount(%v) = %v, want %v", tt.subtotal, got, tt.want)
            }
        })
    }
}

Bounded Autonomy Rules

When generating tests autonomously, follow these rules to decide when to stop and ask the user:

Stop and Ask When

  • Ambiguous requirements — the spec or user story has conflicting or unclear acceptance criteria
  • Missing edge cases — you cannot determine boundary values without domain knowledge (e.g., max allowed transaction amount)
  • Test count exceeds 50 — large test suites need human review before committing; present a summary and ask which areas to prioritize
  • External dependencies unclear — the feature relies on third-party APIs or services with undocumented behavior
  • Security-sensitive logic — authentication, authorization, encryption, or payment flows require human sign-off on test scenarios

Continue Autonomously When

  • Clear spec with numbered acceptance criteria — each criterion maps directly to tests
  • Straightforward CRUD operations — create, read, update, delete with well-defined models
  • Well-defined API contracts — OpenAPI spec or typed interfaces available
  • Pure functions — deterministic input/output with no side effects
  • Existing test patterns — the codebase already has similar tests to follow

Property-Based Testing

Property-based testing generates random inputs to verify invariants instead of relying on hand-picked examples. Use it when the input space is large and the expected behavior can be described as a property.

Python — Hypothesis

from hypothesis import given, strategies as st
from app.serializers import serialize, deserialize

@given(st.text())
def test_roundtrip_serialization(data):
    """Serialization followed by deserialization returns the original."""
    assert deserialize(serialize(data)) == data

@given(st.integers(), st.integers())
def test_addition_is_commutative(a, b):
    assert a + b == b + a

TypeScript — fast-check

import fc from "fast-check";
import { encode, decode } from "./codec";

test("encode/decode roundtrip", () => {
  fc.assert(
    fc.property(fc.string(), (input) => {
      expect(decode(encode(input))).toBe(input);
    })
  );
});

When to Use Property-Based Over Example-Based

Use Property-Based Example
Data transformations Serialize/deserialize roundtrips
Mathematical properties Commutativity, associativity, idempotency
Encoding/decoding Base64, URL encoding, compression
Sorting and filtering Output is sorted, length preserved
Parser correctness Valid input always parses without error

Mutation Testing

Mutation testing modifies your production code (creates "mutants") and checks whether your tests catch the changes. If a mutant survives (tests still pass), your tests have a gap that coverage alone cannot reveal.

Tools

Language Tool Command
TypeScript/JavaScript Stryker npx stryker run
Python mutmut mutmut run --paths-to-mutate=src/
Java PIT mvn org.pitest:pitest-maven:mutationCoverage

Why Mutation Testing Matters

  • 100% line coverage != good tests — coverage tells you code was executed, not that it was verified
  • Catches weak assertions — tests that run code but assert nothing meaningful
  • Finds missing boundary tests — mutants that change < to <= expose off-by-one gaps
  • Quantifiable quality metric — mutation score (% mutants killed) is a stronger signal than coverage %

Recommendation: Run mutation testing on critical paths (auth, payments, data processing) even if overall coverage is high. Target 85%+ mutation score on P0 modules.


Cross-References

Skill Relationship
engineering/spec-driven-workflow Spec → acceptance criteria → test extraction pipeline
engineering-team/focused-fix Phase 5 (Verify) uses TDD to confirm the fix with a regression test
engineering-team/senior-qa Broader QA strategy; TDD is one layer in the test pyramid
engineering-team/code-reviewer Review generated tests for assertion quality and coverage completeness
engineering-team/senior-fullstack Project scaffolders include testing infrastructure compatible with TDD workflows

Limitations

Scope Details
Unit test focus Integration and E2E tests require different patterns
Static analysis Cannot execute tests or measure runtime behavior
Language support Best for TypeScript, JavaScript, Python, Java
Report formats LCOV, JSON, XML only; other formats need conversion
Generated tests Provide scaffolding; require human review for complex logic

When to use other tools:

  • E2E testing: Playwright, Cypress, Selenium
  • Performance testing: k6, JMeter, Locust
  • Security testing: OWASP ZAP, Burp Suite

04 同仓库更多 Skills

GongYuanCaiJi
GongYuanCaiJi/dsh-claude-skills

syllabus

Generates a curated supplementary reading list from any course syllabus using Consensus academic search. Grill-me intake (syllabus input format + course audience + year range) plus a grouping forcing-options checkpoint before any search runs — so the reading list matches the course's level and recency need. Parses the syllabus to extract topics and learning outcomes, searches Consensus for recent peer-reviewed papers per topic, and produces a professionally formatted .docx with clickable Consensus links, plain-language summaries calibrated to audience level, and Bloom-higher-order discussion questions tied to course learning goals. Use when the user uploads a syllabus, course outline, or curriculum document and wants supplementary readings (e.g., 'create a reading list from this syllabus', 'find recent papers for my course') — even casual mentions with a syllabus attached should trigger this skill.

★ 0
GongYuanCaiJi
GongYuanCaiJi/dsh-claude-skills

research

Default entry point for any research request — a hybrid router that classifies the question deterministically and either delegates to a specialist research skill (pulse for trends/sentiment, grants for NIH funding, litreview for academic literature, syllabus for course reading, patent for prior-art + IP landscape, dossier for entity research) or runs its own plan-decompose-multi-source-search-synthesize-cite fallback workflow when no specialist matches. Always surfaces the routing decision so users can override. Use when the user makes any research request that doesn't obviously match a more-specific specialist skill (e.g., "research [topic]", "look into [topic]", "what do we know about [topic]", "investigate [topic]", "find me information on [topic]", "do some research on [topic]", "I need to understand [topic]"). Output is a markdown briefing (default) or .docx document (on request) with full citations and an audit log.

★ 0
GongYuanCaiJi
GongYuanCaiJi/dsh-claude-skills

pulse

Multi-source recency research skill that takes the pulse of any topic across Reddit, Hacker News, the open web, and optionally X/Twitter within a configurable recent window (default 30 days). Forcing intake clarifies topic specificity, angle (trend/sentiment/problems/opportunities/comparison), time window, and platform scope before searching. Returns a synthesized briefing with citations, engagement metrics, and cross-platform pattern analysis. Use when the user requests multi-source recency intelligence on a topic (e.g., 'pulse on [topic]', 'what's happening with [topic]', 'what are people saying about [topic]', 'current conversation about [topic]', 'take the pulse of [topic]', 'trending: [topic]', 'find me info on [topic]'), and for competitor research, trend discovery, tool comparisons, and audience sentiment analysis.

★ 0
GongYuanCaiJi
GongYuanCaiJi/dsh-claude-skills

patent

Patent prior-art and landscape intelligence skill — not generic patent help. Commits to one of five sub-use-cases via forcing intake (novelty search / freedom-to-operate / competitive landscape / acquisition diligence / litigation prior-art) before any search runs. Searches Google Patents, Espacenet, USPTO, and optionally Lens.org for citation-graph signals. Output is an editable Word document (.docx) with verdict, ranked closest art (claim-text extracted), CPC-class-aware landscape, family-resolved hits, geographic coverage, FTO flags where applicable, strategy recommendations, and full audit log. Use when the user asks for patent searching or analysis (e.g., 'prior art search for [invention]', 'freedom to operate analysis for [product]'). Produces search signal, not legal advice — always recommends consulting a patent attorney before filing or licensing decisions. Trademark, copyright, and trade-secret questions are out of scope.

★ 0
GongYuanCaiJi
GongYuanCaiJi/dsh-claude-skills

notebooklm

Browser automation skill for controlling Google's NotebookLM. Use when the user wants anything done in NotebookLM (e.g., 'open NotebookLM', 'check my [name] notebook', 'ask my notebook about X', 'add [source] to NotebookLM', 'generate a Video Overview from my notebook', 'use NotebookLM Studio'). Handles reading and querying notebooks, adding sources (URLs, text, files, YouTube links, synthesized content), generating Studio outputs (Audio/Video Overviews, Mind Maps, Reports incl. Briefing Doc/Study Guide/FAQ, Flashcards, Quiz, slide decks, infographics — discover the exact set from the live Studio panel; the UI evolves fast), and creating new notebooks. Requires browser automation environment — fails gracefully when unavailable.

★ 0
GongYuanCaiJi
GongYuanCaiJi/dsh-claude-skills

litreview

Academic literature orientation skill that searches papers via free keyless APIs (PubMed E-utilities + OpenAlex) by default — with the Consensus MCP as an optional enhancement lane when connected — builds a strategic search plan using PICO (default) or SPIDER / Decomposition / hybrid as fallbacks, and synthesizes findings into a formatted Word (.docx) research guide. Grill-me intake (research question specificity + framework hint + tentative depth) before the recon search; a second forcing checkpoint after Phase 2 confirms framework + sub-areas + depth before searches consume budget. Configurable depth (5/10/20 queries) controls coverage vs. speed. Output is a 'launching pad' — an orientation guide that lets a researcher dive in confidently, not a finished review. Use when the user starts literature-oriented research (e.g., 'litreview on [topic]', 'literature review on [topic]', 'I'm starting a literature review on X', 'I'm writing a paper on X', 'help me research X', 'I'm doing research on X', 'can you help me research X'). Do NOT use for single one-off paper searches wanting a quick list — that's a plain PubMed/OpenAlex (or Consensus) query.

★ 0