revfactory / harness

2026-03-26 · Source: Github Trending: All languages · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Robotics & Autonomous Systems · Depth: Intermediate, medium

Summary

Harness is a Claude Code plugin, currently at Version 1.2.0, designed as a team-architecture factory for generating agent teams and their associated skills. Operating at the L3 Meta-Factory layer, it automates the creation of agent definitions (".claude/agents/") and skills (".claude/skills/") based on a domain description. The plugin offers six pre-defined architectural patterns: Pipeline, Fan-out/Fan-in, Expert Pool, Producer-Reviewer, Supervisor, and Hierarchical Delegation, facilitating complex task decomposition. Key features include agent team design, skill generation with Progressive Disclosure, orchestration, and validation. A controlled experiment demonstrated that using Harness improved average output quality by +60% (from 49.5 to 79.3), achieved a 15/15 win-rate, and reduced output variance by -32% across 15 software engineering tasks. It integrates with other tools like Archon for runtime configurations and has a Codex port, meta-harness.

Key takeaway

For AI Engineers designing complex multi-agent systems within Claude Code, Harness offers a structured approach to automate team architecture and skill generation. If you are struggling with agent coordination or inconsistent outputs, consider integrating Harness to leverage its six architectural patterns. An internal pilot can help you validate the reported +60% average quality improvement and -32% output variance for your specific use cases.

Key insights

Harness automates Claude Code agent team and skill generation using six architectural patterns for complex task decomposition.

Principles

• Decompose complex tasks with specialized agents.
• Structure agent teams using proven architectural patterns.
• Automate skill generation for efficient context management.

Method

Harness follows a six-phase workflow: domain analysis, team architecture design, agent definition generation, skill generation, integration/orchestration, and validation/testing.

In practice

• Use "build a harness for this project" prompt.
• Select from 6 architectural patterns for teams.
• Generate agent definitions and skill files automatically.

Topics

• Claude Code
• Agent Teams
• Multi-Agent Systems
• AI Orchestration
• Architectural Patterns
• Skill Generation

Code references

• revfactory/harness
• coleam00/Archon
• SaehwanPark/meta-harness
• affaan-m/everything-claude-code
• wshobson/agents

Best for: AI Engineer, Machine Learning Engineer, AI Architect

Related on AIssential

• The Sequence Opinion #844: Harness Engineering: The Operating System for Agentic Software — Harness engineering shifts focus from perfect prompts to robust environments for reliable AI agent operation.
• I Trained Claude Code To Run My X Account (no API) — Claude Code can be trained to autonomously navigate and interact with web platforms for complex tasks.
• Extreme Harness Engineering for Token Billionaires: 1M LOC, 1B toks/day, 0% human code, 0% human review — Ryan Lopopolo, OpenAI Frontier & Symphony — Harness engineering enables AI agents to autonomously generate and manage code, accelerating product development.
• LAI #120: Beyond Prompting — Harness engineering focuses on AI system design, routing, and reliability beyond basic prompting.
• Harnessing Claude’s intelligence — Building with evolving AI requires adapting agent harnesses to leverage new model capabilities and optimize performance.
• Components of A Coding Agent — Agentic harnesses significantly enhance LLM coding capabilities by integrating tools, context, and memory management.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Github Trending: All languages.