harness-init

Scaffold a standardized agent-first development harness with AGENTS.md, docs, and state files.

1|Updated Feb 11, 2026
One-click install
npx skills add https://github.com/LauraFlorentin/skills-marketplace --skill harness-init-lauraflorentin
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: harness-init
Source: https://github.com/LauraFlorentin/skills-marketplace/tree/main/harness-engineer/skills/harness-init
Command: npx skills add https://github.com/LauraFlorentin/skills-marketplace --skill harness-init-lauraflorentin

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Scaffold a production-ready harness for agent-first development. Run once per project to orient agents and maintain context across sessions. Creates a standardized project scaffold with TOC, docs, and handoff artifacts.

Core Features & Use Cases

  • Generates AGENTS.md with a concise project map and operating principles.
  • Creates docs/, including architecture, design plans, history, and beliefs to guide agent sessions.
  • Produces features.json to track features and a stable verification contract.
  • Provides init.sh to start a dev server and smoke test, plus runtime state management in current_tasks/.
  • Generates layers.json and a .harness/ structure to enforce architectural constraints.
  • Produces claude-progress.txt as a handoff log for cross-session coordination.

Quick Start

Run the init.sh script at the repo root to scaffold the harness and initialize the project for agent-first development.

Frequently Asked Questions about harness-init

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

FAQPage Schema
How do I scaffold a project for agent-first development?

Scaffolding a project for agent-first development is done by running an initialization script at the repository root to generate a standardized harness with context files and architectural constraints. This setup produces AGENTS.md, docs, and handoff artifacts.

What is a development harness for AI agents?

A development harness for AI agents is a standardized project structure that maintains context across sessions. It includes a project map in AGENTS.md, documentation, feature tracking, and a handoff log to guide autonomous operations.

How do I maintain AI agent context across development sessions?

To maintain AI agent context across sessions, you use a generated harness that includes a progress log and current tasks directory. These artifacts capture session state and handoff information for continuous coordination.

What files do I need for an agent-first project setup?

Files needed for an agent-first project setup include AGENTS.md for the project map, docs for architecture, features.json for verification contracts, init.sh for startup, and layers.json for constraints. A progress log coordinates cross-session work.

Can I use project scaffolding automation for existing repositories?

Project scaffolding automation can be applied to existing repositories or new projects to orient agents. Running the initialization script wires the standardized harness structure into the repository root.

Does project scaffolding require dependencies to generate the harness structure?

Project scaffolding to generate the harness structure requires no external dependencies. The automation runs standalone to create the necessary documentation, state files, and startup script.