AppVerk

Av Marketplace — Testing skill for Claude Code

Testing community

One AI development harness for Claude Code and Oh My Pi (OMP): plugins that plan, deliver task by task with Python, React and PHP agents, test the running app, review with adversarial verification and.

How to install Av Marketplace

This entry records only its repository, not the path inside it, so there is no exact command to give. Open AppVerk/av-marketplace and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Av Marketplace does

Alternatives in Testing

OpenAgentsControl — AI agent framework for plan-first development workflows with approval-based execution 4.8k ★
Deliver — Delivery phase - Review, validate, and test with multi-AI quality assurance 2.8k ★
Spec Driven Develop — Spec-driven development workflow for AI coding agents: architecture-first planning, task decomposition, GitHub 975 ★

README

av-marketplace

**One AI development harness for Claude Code and Oh My Pi.**
Plan, deliver, test, review and commit, with each stage handing its output to the next.

[![MIT License](https://img.shields.io/badge/license-MIT-blue.svg)](LICENSE) [![Plugins](https://img.shields.io/badge/plugins-14-green.svg)](#available-plugins) [![Claude Code](https://img.shields.io/badge/Claude_Code-supported-D97757.svg)](#claude-code) [![Oh My Pi](https://img.shields.io/badge/Oh_My_Pi-supported-A98BFA.svg)](#oh-my-pi-omp)

![From plan to pull request, one flow for both tools. Plan: write the plan in plan mode, or in Claude Code with superpowers, where you brainstorm a spec, check it with /superutils:spec-review and plan from it; in Oh My Pi, Plan Review has a second model review the plan; you approve the plan, which starts the delivery. Delivery, for every task: route it to the agent that owns its files, implement, review with up to 3 fix rounds, commit. After the last task: verification, QA and code review, then you choose which findings to fix, or set after_review = "fix-all" to fix them automatically, and open the PR.](docs/assets/harness-flow.png)

Why use it

Plans run task by task. Delivery sends each task to the developer agent that owns its files, has it reviewed and gives it its own commit.
Every result gets a second look. A task reviewer checks each task's code before it is committed. In every code review a challenger tests the serious findings for false positives and a cross-verifier correlates findings across domains. In Oh My Pi a second model also reviews each plan, and with the recommended role mapping each check runs on a different model family than the work it checks.
QA exercises the running app. /qa:run tests in the browser, through APIs and data stores, and in desktop apps, then can fix the issues you approve and retest within fixed limits.
Findings have IDs. Code review and QA findings carry IDs lik