AI QA agents

Agentic QA means agents that do real testing work — generating, running, and repairing tests — with a human approving what lands. Two are shipped; two are in development.

Shipped

QA-Core

Agentic testing platform

Five-stage agentic pipeline for self-healing Playwright tests, with its deterministic healing engine published on npm as qa-core-heal.

Explore QA-Core →

Test Generator Agent

v1.4
Problem

Turning requirements into structured test cases is the slowest manual step in QA.

How it works

A 7-step generation pipeline with webhook intake via Express, a React review UI, and multi-format export so generated cases drop into existing test management tools.

Stack
Node.jsExpressReactClaude API
Roadmap

APIForge

In development
Problem

API test suites go stale as endpoints evolve, and contract and negative tests are slow to write by hand.

Planned approach

Will generate Playwright API test suites from an API spec — covering happy paths, negative cases, and schema validation — structured for CI.

TypeScriptPlaywrightClaude API

Data Validation Agent

In development
Problem

Data pipelines fail silently; schemas drift and bad data reaches dashboards before anyone notices.

Planned approach

Will run automated source-to-target validation, comparing schemas, row counts, and field-level rules across pipeline stages and reporting the exact failing records.

PythonSQLClaude API
Discuss an agent project →
Muhammad Usman
Open to work
© 2026 Muhammad Usman ·ISTQB® CTFLUpwork Top Rated+10+ Yrs Experience