qa-ai-agents
Enables automated QA testing by running a pipeline of AI agents that generate test scenarios, architect test layers, write Playwright tests, and review code, all grounded in feature requirements and API contracts.
README
QA AI Agents
An AI-powered QA automation pipeline built with TypeScript, Claude API, and the Model Context Protocol (MCP).
Architecture
Four specialised agents run in sequence, each grounded in feature requirements, business rules, and API contracts:
- Agent 1 — Functional Tester: Generates test scenarios across happy path, negative, security, edge cases, UI states, and business rules
- Agent 2 — Test Architect: Assigns each scenario to the correct test pyramid layer (unit / API / component / E2E)
- Agent 3 — Test Engineer: Writes production-grade Playwright tests using Page Object Model (POM)
- Agent 4 — Code Reviewer: Audits generated tests for selector quality, assertion completeness, and Playwright best practices
MCP Server
Exposes the pipeline as 4 MCP tools callable from Claude Desktop or Claude Code:
run_qa_pipeline— full 4-agent pipelinegenerate_test_scenarios— Agent 1 onlyset_feature_input— write to input/feature.mdget_pipeline_report— read last pipeline report
Tech Stack
- TypeScript + Node.js
- Anthropic Claude API (Sonnet 4.6 + Haiku 4.5)
- Ollama (local LLM support — toggle via USE_OLLAMA)
- Playwright (generated test output)
- Model Context Protocol (MCP) SDK
Setup
git clone https://github.com/abchahal/qa-ai-agents.git
cd qa-ai-agents
npm install
cp .env.example .env
# Add your ANTHROPIC_API_KEY to .env
Running the pipeline
# Via terminal
npm run pipeline
# Via Claude Desktop / Claude Code
# Type: "Run the full QA pipeline using input/feature.md"
Input files
| File | Purpose |
|---|---|
input/feature.md |
Feature requirements, business rules, UI selectors |
input/api-contract.json |
API schema (optional) |
input/source.js |
Source code context (optional) |
Output
output/
├── pages/ ← Page Object classes
├── tests/ ← Playwright spec files
└── pipeline_report.md
Model Strategy
| Agent | Model | Reason |
|---|---|---|
| Agent 1 | Haiku 4.5 | Structured JSON output |
| Agent 2 | Haiku 4.5 | Classification task |
| Agent 3 | Haiku 4.5 | Code generation |
| Agent 4 | Haiku 4.5 | Analysis and scoring |
MCP Setup via CLI
Step 1 — Create the batch file
Create start-mcp.bat in the project root:
@echo off
cd /d "C:\path\to\qa-ai-agents"
node --loader ts-node/esm src/server.ts
Replace C:\path\to\qa-ai-agents with your actual project path.
Step 2 — Register the MCP server
claude mcp add -s user qa-ai-agents "C:\path\to\qa-ai-agents\start-mcp.bat"
Step 3 — Verify connection
# List all MCP servers and their status
claude mcp list
# Check your server specifically
claude mcp get qa-ai-agents
Expected output:
qa-ai-agents:
Scope: User config (available in all your projects)
Status: ✔ Connected
Type: stdio
Command: C:\path\to\qa-ai-agents\start-mcp.bat
Step 4 — Remove the server (if needed)
claude mcp remove qa-ai-agents -s user
Switching between Ollama and Claude API
Ollama → Claude API
Step 1 — Update .env:
USE_OLLAMA=false
ANTHROPIC_API_KEY=sk-ant-your-actual-key-here
Step 2 — Restart the MCP server to pick up new env:
claude mcp remove qa-ai-agents -s user
claude mcp add -s user qa-ai-agents "C:\path\to\qa-ai-agents\start-mcp.bat"
Step 3 — Verify correct provider is loaded:
npm run agent1
You should see:
Using Claude API: claude-sonnet-4-6
Claude API → Ollama
Step 1 — Make sure Ollama is running and model is pulled:
ollama list
# Should show qwen2.5-coder:7b or your preferred model
If model is not pulled yet:
ollama pull qwen2.5-coder:7b
Step 2 — Update .env:
USE_OLLAMA=true
OLLAMA_MODEL=qwen2.5-coder:7b
Step 3 — Restart the MCP server:
claude mcp remove qa-ai-agents -s user
claude mcp add -s user qa-ai-agents "C:\path\to\qa-ai-agents\start-mcp.bat"
Step 4 — Verify correct provider is loaded:
npm run agent1
You should see:
Using Ollama local model: qwen2.5-coder:7b
Provider comparison
| Ollama (local) | Claude API (cloud) | |
|---|---|---|
| Cost | Free | ~$0.18 per pipeline run |
| Speed | 15–25 minutes | 45–90 seconds |
| Quality | Good | Best |
| Internet required | No | Yes |
| Best for | Development and debugging | Production runs and demos |
Tip: Use
USE_OLLAMA=truewhile writing and debugging agent code to avoid burning API credits. Switch toUSE_OLLAMA=falsefor actual pipeline runs and portfolio demos.
Environment variables
| Variable | Required | Description |
|---|---|---|
ANTHROPIC_API_KEY |
Yes (if USE_OLLAMA=false) | Your Anthropic API key from console.anthropic.com |
USE_OLLAMA |
Yes | true for local Ollama, false for Claude API |
OLLAMA_MODEL |
No | Ollama model name. Default: qwen2.5-coder:7b |
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.