harness

Name: harness
Availability: InStock
Author: junnv93

Orchestrate a Planner, Generator, and Evaluator loop for multi-file implementation tasks.

Updated Aug 23, 2026

Search Tags:#verification #orchestration #generator #ai-agents #planner #harness #evaluator

Ready to run

npx skills add https://github.com/junnv93/equipment_management_system --skill harness-junnv93

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: harness
Source: https://github.com/junnv93/equipment_management_system/tree/main/.claude/skills/harness
Command: npx skills add https://github.com/junnv93/equipment_management_system --skill harness-junnv93

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Harness coordinates a Planner → Generator → Evaluator loop to tackle multi-file tasks by auto-selecting execution modes and reusing existing verify-/review- infrastructure as Evaluator components.

Core Features & Use Cases

Reuses existing verify-* and review-* as Evaluator infrastructure to avoid reinventing verification loops.
Auto-selects execution mode based on task complexity to optimize resource use.
Maintains structured, auditable handoffs between Planner, Generator, and Evaluator with exec-plan-style coordination.

Quick Start

Activate Harness to start the Planner → Generator → Evaluator loop and monitor the first iteration.

Frequently Asked Questions about harness

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

FAQPage Schema

How do I orchestrate an automated planner generator evaluator loop for complex software tasks?▼

An automated planner generator evaluator loop coordinates structured handoffs to validate complex software tasks. It auto-selects execution mode based on task complexity and reuses existing verification infrastructure to guide automated iteration toward a final validated result.

What is the best way to validate multi-file implementation tasks automatically?▼

Validating multi-file implementation tasks automatically requires an orchestration loop that manages structured handoffs between generation and evaluation. This approach stores execution plans and evaluation records to ensure automated iteration reaches a final validated result.

Can I reuse existing verification and review skills as evaluator infrastructure for AI agents?▼

You can reuse existing verify-* and review-* skills as evaluator infrastructure for AI agents. This avoids reinventing verification loops by integrating these existing skills directly into the automated evaluation phase of the orchestration cycle.

How does task complexity auto-selection work for AI agent orchestration?▼

AI agent orchestration auto-selection works by evaluating task complexity to determine the optimal execution mode. This optimizes resource use by matching the execution approach to the specific demands of the multi-file implementation task.

How do I maintain auditable handoffs between an AI planner, generator, and evaluator?▼

Auditable handoffs between a planner, generator, and evaluator are maintained using exec-plan-style coordination. The system stores execution plans and evaluation records throughout the automated iteration loop to ensure structured and traceable transitions.

When do I need an automated planner generator evaluator loop instead of manual code verification?▼

An automated planner generator evaluator loop is needed for complex multi-file implementation tasks where manual verification is insufficient. It enforces structured handoffs and reuses verification infrastructure to guide automated iteration toward a validated result.