AgentCommunity★ 1k

Sub-Agent Collective

Updated Jul 7, 2026

An NPX-installable collection of 30+ TDD-focused Claude Code agents by vanzan01. A central /van command routes work through @task-orchestrator to the right specialist. All agents enforce RED→GREEN→REFACTOR — tests are written before implementation. Context7 integration pulls real library documentation instead of letting agents guess at APIs.

View on GitHub

What it does

/van
Routes a task through @task-orchestrator, which selects the right specialist and enforces TDD from the start.
@task-orchestrator
Central routing hub that analyzes request complexity and delegates to the appropriate specialist agent.
@feature-implementation-agent
Implements business logic with comprehensive test coverage, enforcing RED→GREEN→REFACTOR before handoff.
@research-agent
Pulls real, current library documentation via Context7 before agents write code against an API.
@functional-testing-agent
Runs browser-based end-to-end tests with Playwright as part of the validation pipeline.
npx claude-code-collective validate
Validates that all agents, hooks, and CLAUDE.md behavioral rules are correctly installed.

Files (1)

OVERVIEW.mdprimary · markdown · 2.0 KB

## What it does

An NPX-installable collection of 30+ Claude Code agents centered on TDD and Context7-powered documentation lookup. The /van entry point routes all requests through @task-orchestrator, which analyzes complexity and picks the right specialist. Every implementation agent delivers a TDD completion contract showing RED, GREEN, and REFACTOR phases with test results before handing off.

Core implementation agents: @component-implementation-agent (UI components with tests), @feature-implementation-agent (business logic with coverage), @infrastructure-implementation-agent (build systems), @testing-implementation-agent (test suites), @polish-implementation-agent (performance optimization).

Quality and validation: @quality-agent (code review), @devops-agent (CI/CD), @functional-testing-agent (Playwright browser testing), @enhanced-quality-gate, @completion-gate.

Research: @research-agent (Context7 documentation), @prd-research-agent (requirement breakdown).

Installation installs CLAUDE.md behavioral rules, .claude/settings.json with hook configuration, .claude/agents/ (30+ agent files), .claude/hooks/ (test-driven-handoff.sh, collective-metrics.sh), and .claude-collective/ (Vitest test templates, metrics).

## Install

    npx claude-code-collective init

For selective install:

    npx claude-code-collective init --minimal       # core agents only
    npx claude-code-collective init --testing-only  # testing framework only
    npx claude-code-collective init --hooks-only    # hooks and behavioral system only
    npx claude-code-collective init --interactive   # guided setup

Restart Claude Code after install — hooks require a fresh session to load.

Validate the install:

    npx claude-code-collective validate
    npx claude-code-collective status

## When to use

When you want TDD enforced automatically rather than by convention. The hook system blocks handoffs until tests exist and pass, so agents cannot skip the RED phase. Works best for greenfield features and MVP prototyping where test discipline matters from the start.