harness

Generate domain-specific agent teams and skills under .claude/agents/ and .claude/skills/.

40|9|Updated Jan 4, 2026
One-click install
npx skills add https://github.com/akillness/oh-my-skills --skill harness-akillness
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: harness
Source: https://github.com/akillness/oh-my-skills/tree/main/.agent-skills/harness
Command: npx skills add https://github.com/akillness/oh-my-skills --skill harness-akillness

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires bash, git, curl, npx, node, and includes scripts (resource) and references (resource) components.

What problem does it solve?

Harness automates the design of domain-specific agent teams and their skills, removing manual scaffolding and enabling repeatable, auditable orchestration.

Core Features & Use Cases

  • Design domain-specific agent teams using the 6 architectural patterns (Pipeline, Fan-out/Fan-in, Expert Pool, Producer-Reviewer, Supervisor, Hierarchical Delegation) to match task structure.
  • Auto-generate and bundle skills with progressive disclosure, including SKILL.md scaffolding and references for deep dives.
  • Produce .claude/agents/ and .claude/skills/ artifacts, plus orchestration templates and validation hooks for end-to-end workflows.
  • Provide a general orchestration framework applicable to research, coding, data pipelines, content creation, and product development, with deterministic execution and robust error handling.
  • Support validation and testing workflows (trigger evals, dry-run tests, and baseline vs with-skill comparisons) to ensure reliability.

Quick Start

Trigger harness with prompts like "build a harness for this project" to generate domain-specific agent teams and skills.

Frequently Asked Questions about harness

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I design domain-specific agent teams for automated task orchestration?

Design domain-specific agent teams by selecting an orchestration pattern like pipeline, fan-out/fan-in, or expert pool, then auto-generating the corresponding .claude/ agents and skills artifacts with validation loops for deterministic execution.

What orchestration patterns can I use to structure multi-agent workflows?

Multi-agent workflows can be structured using six orchestration patterns: Pipeline, Fan-out/Fan-in, Expert Pool, Producer-Reviewer, Supervisor, and Hierarchical Delegation, each matching different task structures for reliable integration.

How do I auto-generate claude-code skills and agent definitions for a project?

Auto-generate claude-code skills and agent definitions by triggering the harness with a prompt like "build a harness for this project," which scaffolds SKILL.md files and references under the .claude/ directory.

Do I need bash and node installed to run agent team orchestration and skill generation?

Yes, bash, git, curl, npx, and node are required dependencies to execute the orchestration framework, run validation hooks, and generate the agent and skill artifacts deterministically.

How do I validate multi-agent orchestration patterns to ensure deterministic results?

Validate orchestration patterns by defining trigger evals, running dry-run tests, and comparing baseline versus with-skill execution to ensure deterministic results and robust error handling.

What's the best way to structure specialized agents for a data pipeline or research workflow?

Structure specialized agents by matching the task to an orchestration pattern like Fan-out/Fan-in for data pipelines or Expert Pool for research, then generating the bundled skills and validation hooks for end-to-end workflows.