harness-generator

Generate agent harnesses with linters, style enforcement, and plan gates.

3|Updated May 9, 2026
One-click install
npx skills add https://github.com/gcuder/poise --skill harness-generator-gcuder
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: harness-generator
Source: https://github.com/gcuder/poise/tree/main/poise
Command: npx skills add https://github.com/gcuder/poise --skill harness-generator-gcuder

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

Coding agents often lack repository context, leading to drift, architectural violations, and inconsistent code quality. This skill automates the creation of a robust harness that enforces structural boundaries, style invariants, and plan-before-edit workflows.

Core Features & Use Cases

  • Architectural Guardrails: Automatically generates linters and tests that enforce layer models and restrict forbidden packages.
  • Plan Enforcement: Implements a plan gate that prevents uncoordinated edits, ensuring agents follow a defined execution strategy.
  • Session Briefing: Injects non-negotiable rules and active plans into the agent's context at the start of every session.

Quick Start

Ask your coding agent to generate a complete harness for the current repository.

Frequently Asked Questions about harness-generator

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

FAQPage Schema
How do I enforce architectural boundaries for coding agents in my repository?

To enforce architectural boundaries for coding agents, you need a harness that scans your repository to infer layer models and generates automated linters. This skill creates those guardrails to prevent dependency violations and structural drift during automated edits.

What is a plan gate mechanism for agent workflows?

A plan gate mechanism for agent workflows prevents uncoordinated code edits by requiring a defined execution strategy before changes are made. This skill implements plan enforcement to ensure coding agents follow approved workflows rather than making spontaneous modifications.

Can I generate linters for Claude Code, Codex, and OpenCode simultaneously?

Yes, you can generate linters for Claude Code, Codex, and OpenCode. This skill creates a comprehensive agent harness that operates across these platforms, producing architectural validation scripts and style enforcement rules tailored to your specific codebase structure.

Do I need Python 3 to generate agent harnesses and run architectural validation?

Yes, Python 3 is required to execute the generated quality checks and architectural validation scripts. You also need standard language-specific linters installed in your environment to run the harness rules generated for your codebase.

How do I automatically inject coding rules and active plans into an agent session?

To automatically inject coding rules and active plans into an agent session, this skill generates a session briefing mechanism. It injects non-negotiable rules and active plans into the agent context at the start of every session to maintain code quality.

What's the best way to prevent code quality drift when using automated coding agents?

The best way to prevent code quality drift with automated coding agents is to implement a generated harness with style enforcement and structural linters. This skill scans your repository structure to automatically generate these guardrails and plan-gate mechanisms.