revfactory/harness

A meta-skill that designs domain-specific agent teams, defines specialized agents, and generates the skills they use.

What it solves

Harness solves the difficulty of manually designing and configuring complex multi-agent teams for Claude Code. Instead of writing individual agent definitions and skills by hand, users can describe their domain in a simple sentence, and the tool automatically generates the necessary architecture and skill sets to handle complex tasks.

How it works

Harness acts as a "Team-Architecture Factory." When a user provides a domain description, it analyzes the requirements and applies one of six pre-defined architectural patterns to generate agent definitions (.claude/agents/) and skills (.claude/skills/).

It supports two execution modes:

  • Agent Teams: For tasks requiring collaboration between two or more agents using coordination protocols.
  • Subagents: For one-off tasks that do not require inter-agent communication.

Who it’s for

Developers and users of Claude Code who want to deploy specialized, coordinated teams of AI agents to handle complex workflows like deep research, full-stack development, content planning, or technical documentation without manual configuration.

Highlights

  • Six Architecture Patterns: Includes Pipeline, Fan-out/Fan-in, Expert Pool, Producer-Reviewer, Supervisor, and Hierarchical Delegation.
  • Automated Generation: Turns domain descriptions into fully functional agent teams and skills.
  • Context Management: Uses Progressive Disclosure for efficient skill generation.
  • Validation Tools: Includes trigger verification, dry-run testing, and comparison tests to evaluate performance with and without skills.

Related

  • Project
    shareAI-lab/learn-claude-code

    A 17‑lesson tutorial that teaches you how to build the surrounding “harness” code for an LLM‑based coding agent (Claude). It walks through adding tools, permissions, context management, memory, task scheduling, multi‑agent coordination, and goal‑driven termination, all around a simple agent‑loop that lets the model decide when to call tools.

  • Project
    wshobson/agents

    A marketplace of production-ready agentic building blocks including plugins, agents, and skills that can be deployed across multiple AI coding harnesses like Claude Code and Cursor.

  • Project
    rsmdt/the-startup

    A multi-agent AI framework for Claude Code that implements a spec-driven development workflow to turn ideas into shipped code via structured planning and parallel specialist agents.

  • Project
    ruvnet/metaharness

    A factory for creating custom, repo-aware AI agent harnesses that can be published as standalone npm packages and deployed across multiple agent hosts.

  • Dispatch
    Scaling AI Coding: How Claude Code Navigates Large Codebases

    An exploration of agentic search and the 'harness' architecture used by Claude Code to manage multi-million line monorepos and legacy systems without relying on traditional RAG indexing.