skills.rest
enkira-ai
enkira-aiOFFICIAL
Explore all skills by this author →
WebsiteGitHub

harness-verify

Automates independent feature verification in harness-engineered repositories using Codex, Gemini, or Claude subagent fallback.

2|Updated Mar 7, 2026
Search Tags:#workflow#verification#codex#gemini#harness#independent-evaluation
One-click install
npx skills add https://github.com/enkira-ai/claude-plugins --skill harness-verify-enkira-ai
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: harness-verify
Source: https://github.com/enkira-ai/claude-plugins/tree/main/plugins/harness-engineer/skills/harness-verify
Command: npx skills add https://github.com/enkira-ai/claude-plugins --skill harness-verify-enkira-ai

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Automates independent verification of completed features in harness-engineered repos by ensuring unbiased evaluation using a separate coding CLI.

Core Features & Use Cases

  • Verifier selection: Use codex, gemini, or a Claude subagent as fallback to isolate the evaluator.
  • Protocol-driven verification: automatically locate the feature, detect the verifier, build an evaluator prompt, and parse the verdict.
  • End-to-end safety: guarantees no shared state with the feature writer to prevent bias.

Quick Start

Run harness-verify with the feature ID to begin an independent verification pass.

Frequently Asked Questions about harness-verify

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

FAQPage Schema
How do I verify a completed feature in a harness-engineered repository independently?▼

The harness-verify protocol automates independent feature verification by locating the feature, selecting an external verifier like codex or gemini, building an evaluator prompt, and parsing the final verdict. This protocol-driven process ensures unbiased evaluation without shared state.

Can I use gemini or codex as a verifier for isolated feature evaluation?▼

Yes, you can use codex or gemini as external coding CLI verifiers for isolated feature evaluation. In headless environments where those CLIs are unavailable, a Claude subagent is used as a fallback to maintain isolation from the feature generator.

Why do I need an independent evaluator context for feature verification?▼

An independent evaluator context is needed to guarantee end-to-end safety and prevent bias during feature verification. By isolating the evaluator from the generator with no shared state, you ensure the final verdict accurately reflects the feature's correctness without generator influence.

What's the best way to automate unbiased feature verification in a headless environment?▼

Automating unbiased feature verification in a headless environment is best handled by using a Claude subagent fallback when external CLIs like codex or gemini are unavailable. This maintains the required isolation from the generator while parsing the final verdict.

Does harness-verify require an external coding CLI to parse a final verdict?▼

Harness-verify does not strictly require an external coding CLI to parse a final verdict, as a Claude subagent fallback is supported for headless environments. However, using codex or gemini as verifiers provides the primary isolated context for evaluating the completed feature.

Related Skills

Explore More Software Engineering Skills→
alexanderopalexanderop

using-git-worktrees

Create isolated git worktrees for feature branches with safety checks.

Community
Advanced
m2ai-portfoliom2ai-portfolio
4

build-spec-generator

Generate a deployment-ready build spec from a workflow verdict.

Official
Intermediate
enkira-aienkira-ai
2

harness-work

Automate end-to-end feature work in harness-engineered repositories.

Official
Advanced
hyperb1isshyperb1iss
47

sibyl

Store coding patterns, learnings, and tasks in a knowledge graph.

Community
Advanced