harness-extend

Extend Claude Code harnesses with coordinated phases, scopes, hooks, and commands.

Updated Aug 27, 2026
One-click install
npx skills add https://github.com/ElvinOuyang/claude-skill-collection --skill harness-extend
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: harness-extend
Source: https://github.com/ElvinOuyang/claude-skill-collection/tree/main/plugins/harness-engineering-toolkit/skills/harness-extend
Command: npx skills add https://github.com/ElvinOuyang/claude-skill-collection --skill harness-extend

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It reduces the risk and effort involved in extending an existing Claude Code harness, where changes must be coordinated across multiple files (state initialization, stop-gate requirements, scope evaluation, and command wiring) to avoid deadlocks and inconsistent enforcement.

Core Features & Use Cases

  • Phase extension with consistency guardrails: Adds new enforcement phases to the harness state machine and ensures the phase name is updated everywhere it is checked and marked.
  • Scope addition with correct inference ordering: Introduces new scopes, including the file-pattern trigger and REQUIRED phase mapping, while maintaining the stop gate’s top-to-bottom specificity order.
  • Hook and command expansion: Adds new hook scripts (SessionStart, PreToolUse, PostToolUse, Stop) and/or slash commands, then updates wiring and harness documentation so the new enforcement behavior actually runs.
  • Multi-file coordination checklist: Provides exhaustive update steps to modify all required harness touchpoints together rather than piecemeal.

Quick Start

Ask the skill to extend your harness by adding a new scope and required phases, then follow its checklist to update the stop gate, scope evaluation, tests/smoke logic, and harness documentation.

Frequently Asked Questions about harness-extend

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

FAQPage Schema
How do I add new enforcement phases to a Claude Code harness without causing deadlocks?

Adding new enforcement phases to a Claude Code harness requires coordinating updates across state initialization, stop-gate mappings, and scope evaluation to prevent deadlocks and inconsistent phase marking.

What is the best way to introduce new scopes into a Claude Code harness state machine?

Introducing new scopes into a Claude Code harness state machine requires adding the file-pattern trigger and REQUIRED phase mapping while maintaining the stop gate’s top-to-bottom specificity order for correct inference.

How do you wire new hook scripts and slash commands into an existing harness?

Wiring new hook scripts and slash commands into an existing harness involves adding the scripts and then updating hook wiring settings and harness documentation so the new enforcement behavior actually runs.

Why do Claude Code harness updates cause inconsistent enforcement behavior across files?

Inconsistent enforcement behavior occurs when Claude Code harness updates are applied piecemeal rather than coordinating modifications across state initialization, stop-gate requirements, scope evaluation, and command wiring simultaneously.

Do I need to update harness documentation when adding new PreToolUse or PostToolUse hook scripts?

You need to update harness documentation when adding new PreToolUse or PostToolUse hook scripts to ensure the new enforcement behavior actually runs and remains coordinated with existing stop-gate mappings.