harness

What it is

Harness is a team-architecture factory for Claude Code. It designs agent teams using six architectural patterns and auto-generates skills, producing files under .claude/agents/ and .claude/skills/ tailored to a domain.

How it works

The workflow analyzes a domain, designs an agent team, generates agent definitions and skills, and provides orchestration and validation support. It supports six patterns: Pipeline, Fan-out/Fan-in, Expert Pool, Producer-Reviewer, Supervisor, and Hierarchical Delegation. It includes an orchestration framework, and verification/testing capabilities.

Getting started

Via Marketplace

/plugin marketplace add revfactory/harness

/plugin install harness@harness-marketplace

Direct Installation as Global Skill

# Copy the skills directory to ~/.claude/skills/harness/
cp -r skills/harness ~/.claude/skills/harness

Usage

Trigger in Claude Code with prompts like:

Build a harness for this project
Design an agent team for this domain
Set up a harness

Execution modes:

• Agent Teams (default): TeamCreate + SendMessage + TaskCreate
• Subagents: Direct Agent tool invocation

Architecture patterns are:

• Pipeline
• Fan-out/Fan-in
• Expert Pool
• Producer-Reviewer
• Supervisor
• Hierarchical Delegation

Files generated by Harness:

your-project/
├── .claude/
│   ├── agents/          # Agent definition files
│   │   ├── analyst.md
│   │   ├── builder.md
│   │   └── qa.md
│   └── skills/          # Skill files
│       ├── analyze/
│       │   └── SKILL.md
│       └── build/
│           ├── SKILL.md
│           └── references/

Output

Harness generates domain-specific agent definitions and skills in the .claude directory, ready for integration and orchestration.

Coexistence

Harness sits at the L3 Meta-Factory / Team-Architecture Factory layer and focuses on generating agent teams and their skills, complementing runtime determinism tools like Archon.

Built with Harness

License

Apache 2.0