Claude Code plugin
Test Kitchen
Parallel exploration of implementation approaches - implements multiple variants simultaneously and lets tests determine the winner
Install
/plugin marketplace add 2389-research/claude-pluginsFound in the project README. Check it for setup steps after installing.
git clone https://github.com/2389-research/test-kitchen.gitFrom the README
Test Kitchen Test Kitchen intercepts every build/create/implement request and offers parallel exploration before any code is written. At the design phase, omakase-off offers design exploration (3-5 approaches in parallel). At the implementation phase, cookoff runs parallel implementation competition (2-5 agents each creating their own plan). Tests and the judge pick the winner. Read more: Cookoff: Same Spec, Different Code · Omakase: Show Me Parallel implementation framework with two gate skills: Installation Flow Quick examples Entry gate: omakase-off When you make a build/create request: Option 1 (Brainstorm) proceeds to brainstorming skill (or fallback brainstorming), with passive slot detection. Option 2 (Omakase) does the following: - Quick context gathering (1-2 questions) - Generate 3-5 architectural approaches - Implement ALL in parallel via git worktrees - Run same scenario tests on each - Fresh-eyes review survivors - Present comparison, user picks winner During brainstorming, slot detection can also trigger it: Exit gate: cookoff Option 1 (Cookoff): - Each agent reads the same design doc - Each agent creates their OWN implementation plan - All implement in parallel - Fresh-eyes review, compare results - User picks winner Options 2/3 -- single agent or local implementation proceeds normally. Why aggressive triggers Skills can't passively detect "uncertainty" or "readiness" -- they must claim specific moments in the conversation flow. - Omakase-off claims the BUILD/CREATE moment (before brainstorming) - Cookoff claims the IMPLEMENT moment (after design) Dependencies Test Kitchen orchestrates these skills (falls back gracefully if not installed): - superpowers:brainstorming - superpowers:writing-plans - superpowers:executing-plans -…
How plugins work
- A plugin bundles slash commands, subagents, skills, hooks, MCP servers and status lines so they install together.
- Add the marketplace inside Claude Code with /plugin marketplace add 2389-research/test-kitchen, then run /plugin to pick what to install.
- You can turn a plugin off again from /plugin without deleting anything.
Details
- Repository
- 2389-research/test-kitchen
- License
- MIT
- Last updated
- 25 September 2026
- Topics
- ai-agentsanthropicautomationclaudeclaude-codeclaude-code-skillscomparisonexplorationimplementationparalleltestingvariants
- Listed in
- awesome-claude-code-toolkit
FAQ
What is Test Kitchen?
Parallel exploration of implementation approaches - implements multiple variants simultaneously and lets tests determine the winner It is a plugin for Claude Code, from 2389-research/test-kitchen on GitHub.
How do I install Test Kitchen?
Run /plugin marketplace add 2389-research/claude-plugins inside Claude Code, then follow the setup in the README.
Is Test Kitchen popular?
2389-research/test-kitchen has 5 GitHub stars and 1 forks, and is included in 1 curated Claude Code list. It was last updated on 25 September 2026.
More plugins & marketplaces
Magic CC Codex Worker
5wenqingyu/magic-cc-codex-worker
Parallel OpenAI Codex agents inside Claude Code — multi-agent orchestration with git worktree isolation, resumable sessions, and dual-model PR review. MCP plugin.
Burnd
5garvitsurana271/burnd
Find what's burning a hole in your AI coding budget. Local-first cost-control for Claude Code.
Claude Inject Idle Time
5clankercode/claude-inject-idle-time
Claude Code plugin that injects hidden timing context into every user prompt. 42 tokens, local time with UTC offset, idle seconds since last reply, previous turn duration.
Claude Snapshot
4adhenawer/claude-snapshot
Portable Claude Code setup snapshots — export, diff, and apply your config across machines
Claude Agentic Coding Playbook
5john-wilmes/claude-agentic-coding-playbook
Evidence-based practices for LLM-assisted development. Install scripts, Claude Code configuration, and a research-backed best practices guide with 34 verified citations.
Paco Framework
4PenguinAlleyApps/paco-framework
1+ Markdown-first multi-agent OS for Claude Code. Coordinates 3-16 specialized AI agents (Engineering, QA, Growth, Finance) with file-based dispatch, institutional memory, CEO Gate approval, and 24/7 scheduling. No Python required. MIT
Browse the most popular Claude Code mods, all plugins & marketplaces, or the skills and MCP server directories.