Does UltraQA use different AI agents?

Yes, it uses a multi-agent approach: a qa-tester for verification, an architect (using Opus) for failure diagnosis, and an executor for applying fixes.

Can I stop an active UltraQA cycle?

Yes, you can cancel the process at any time using the /oh-my-claudecode:cancel-ultraqa command, which clears the session state.

What types of verification can UltraQA perform?

It supports standard commands like npm test, build, lint, and typecheck, as well as custom patterns and interactive CLI service testing.

How does UltraQA handle persistent errors?

To save time and tokens, UltraQA will automatically stop if it detects the same failure three times or if it reaches a maximum of five cycles without success.

What is the UltraQA skill for Claude Code?

UltraQA is an autonomous workflow that cycles through testing, diagnosing, and fixing code until a predefined goal (like passing tests or successful builds) is achieved.

UltraQA: Autonomous Testing & Fixing

Name: UltraQA: Autonomous Testing & Fixing
Author: Yeachan-Heo

byYeachan-Heo

•

2,142

Security & Testing

Automates the cycle of running tests, diagnosing failures with an architect agent, and applying code fixes until quality goals are met.

About

UltraQA is a sophisticated multi-agent workflow designed for Claude Code that achieves 'green builds' through autonomous iteration. It initiates a continuous loop consisting of testing (unit tests, builds, linting, or type-checking), failure diagnosis by a high-reasoning architect agent, and precise code execution to resolve issues. This skill is particularly effective for resolving complex regressions, completing large refactors, or ensuring codebase stability without manual intervention, supporting up to five iterative cycles with built-in state tracking and early-exit logic for persistent errors.

Key Features

Intelligent failure diagnosis powered by high-reasoning models
2,142 GitHub stars
Support for multiple goal types including tests, builds, linting, and TypeScript checks
Real-time state tracking and observability via .omc/ultraqa-state.json
Autonomous multi-agent orchestration (Tester, Architect, and Executor agents)
Pattern recognition to prevent infinite loops and excessive token usage

Use Cases

Fixing broken test suites automatically after a major refactoring session
Resolving persistent TypeScript or linting errors across a large codebase
Ensuring project build success and dependency health before deployment

About

Key Features

Intelligent failure diagnosis powered by high-reasoning models
2,142 GitHub stars
Support for multiple goal types including tests, builds, linting, and TypeScript checks
Real-time state tracking and observability via .omc/ultraqa-state.json
Autonomous multi-agent orchestration (Tester, Architect, and Executor agents)
Pattern recognition to prevent infinite loops and excessive token usage

Use Cases

Fixing broken test suites automatically after a major refactoring session
Resolving persistent TypeScript or linting errors across a large codebase
Ensuring project build success and dependency health before deployment