Test-Driven Development Workflow
Prove intended behavior with a failing test before implementing, refactoring, and verifying the change.
The skill explicitly treats browser content as untrusted, forbids accessing cookies, local-storage tokens, and credentials, and restricts navigation based on page content. However, it does not require user confirmation, least-privilege permissions, external-side-effect disclosure, dependency security review, or rollback for general code changes, so points are deducted.
Red-Green-Refactor, test classification, isolation guidance, anti-patterns, and a verification checklist provide a coherent workflow with some failure feedback. However, the environment, test framework, generic npm test command, and referenced supporting files are not reproducible from the supplied material; under static calibration the score is capped below 10.
Triggers, non-fit cases, browser scenarios, and the test decision guide are reasonably clear for logic, bug, and behavior changes. Chinese-language support, cross-stack adaptation boundaries, and fallback behavior when Chrome DevTools MCP or other tooling is unavailable are unspecified, so points are deducted.
The document has clear progressive structure with overview, triggers, workflow, examples, anti-patterns, red flags, and verification. The repository supplies MIT licensing and installation context, but the selected skill lacks its own versioning, changelog, maintenance owner, and update path; referenced material is also not included here, so points are deducted.
The skill directly supplies an actionable test-first, bug-reproduction, and regression-verification workflow. No representative execution result or directly verified output is provided, and the fixed testing ratios and prescriptive rules have limited demonstrated marginal benefit across project types; the static ceiling of 7 applies and uncertainty reduces the score.
A real CI workflow, validation scripts, and evaluation commands provide some auditable evidence. However, the supplied fixture does not directly cover this skill, there is no dedicated key-path test or third-party execution evidence for it, and nothing was executed during this review, so the static ceiling of 5 applies.
- The generic npm test command is not tied to a specific project or test framework and may not run as written.
- Chrome DevTools MCP, browser-testing-with-devtools, and references/testing-patterns.md are referenced but not supplied; their availability and completeness must be confirmed after installation.
- The skill does not specify confirmation, external-effect disclosure, dependency auditing, or rollback procedures around code changes.
- The fixed 80/15/5 test-pyramid ratios may not fit every codebase.
What does this skill do, and when should you use it?
This is the test-driven development skill in the agent-skills repository, intended for implementing logic, fixing bugs, and changing behavior. It requires a failing test first, the minimum implementation needed to pass it, and refactoring while keeping tests green. It covers unit, integration, and end-to-end test allocation, along with state-based assertions, isolation, descriptive naming, and restrained mocking. Browser changes should also receive runtime verification with Chrome DevTools MCP.
Instructs an agent to write a failing test before behavioral changes; create a reproduction test for reported bugs; select unit, integration, or end-to-end tests based on side effects and boundaries; use Arrange-Act-Assert, state-based assertions, DAMP tests, and descriptive names; prefer real implementations, then fakes and stubs, while using mocks sparingly; run tests after refactoring; verify completed implementations with npm test and checks for new behavior coverage, regression tests, skipped tests, and coverage changes; for browser work, use Chrome DevTools MCP to inspect the console, network, DOM, styles, performance, and screenshots.
- A developer is implementing new business logic and needs executable proof of the intended behavior.
- A developer receives a bug report and wants to confirm the defect before changing production code.
- A team modifies existing functionality or adds edge-case handling while preserving current behavior.
- A frontend developer changes browser UI or interactions and needs both automated and runtime verification.
- A maintainer refactors working code without changing its externally observable behavior.
What are this skill's strengths and limitations?
- Provides a concrete Red-Green-Refactor cycle and completion checklist.
- Covers bug-reproduction tests, the test pyramid, and resource-based test sizing.
- Emphasizes testing behavior rather than implementation details and defines when mocks are justified.
- Includes browser runtime verification guidance and browser security boundaries.
- Assumes the project has a runnable test environment but does not configure a particular test framework.
- The README names npm test as a verification command without documenting the target project's script configuration.
- Browser verification depends on Chrome DevTools MCP, which may not be available in every client.
- The supplied source contains no results from testing this skill in a specific codebase or platform.
How do you install this skill?
The repository README provides an individual-install command through the skills CLI: npx skills add addyosmani/agent-skills --skill test-driven-development. You can browse the collection first with npx skills add addyosmani/agent-skills --list. The repository also documents integrations for Claude Code, Codex, Cursor, Antigravity CLI, Gemini CLI, Windsurf, OpenCode, GitHub Copilot, and Kiro; separate installation details for this skill are not otherwise specified.
How do you use this skill?
Before implementing logic, fixing a bug, changing existing behavior, or adding edge-case handling, ask the agent to apply the test-driven-development workflow, for example: “Use test-driven development: write a failing test first, implement the minimal fix, then refactor and run the tests.” For browser changes, follow the reproduce, inspect, diagnose, fix, and verify workflow with Chrome DevTools MCP. Pure configuration, documentation, and behavior-neutral static-content changes are out of scope.
How does this skill compare with similar options?
The README names Superpowers and Matt Pocock's skills as comparison targets and links to a comparison document. The supplied material does not include that document's conclusions, so it does not support a feature or quality ranking.