$sqazero-touch-agentMCP SERVER

Give your AI a URL. Get a tested app back.

An MCP server that plugs into Claude, ChatGPT, Cursor, Windsurf, and any AI you already use. Your AI calls SimplifyQA; SimplifyQA calls your app. You provide two things, a URL and login credentials, and 50+ MCP tools per technology handle the rest: exploration, test-case generation, execution, and defect reporting across all 9 technologies.

setup in 60s
2
Things to provide (URL + creds)
9
Technologies covered
50+
MCP tools per technology
0
AI vendor lock-in
SimplifyQAMCP
50+ tools · per tech · online
Claude
ChatGPT
Cursor
Cursor
Windsurf
WEBMOBILEAPIDESKTOPDBMAINSAPHYBRIDGENERIC
Target technologies
MCP · ONLINE
TOOLS · 50+ /TECH
LATENCY 42MS
TARGETS · 9
Claude Desktop · MCP: sqa-zero-touch-agent connected
YYou
Test https://shop.acme.com with admin@acme.com / hunter2. Save to project ACME Shop. Run the suite.
CClaude · sqa-agent
✓[explore]walked 24 pages, classified checkout, cart, browse, login, profile
✓[generate]41 test cases across 5 flows
✓[save]suite ACME_Checkout_v1 in SimplifyQA / Payments
✓[execute]staging: 37 passed, 3 failed, 1 self-healed
✓[defects]filed in Jira: ACMESH-1041, 1042, 1043
✓[rtm]exported to Excel
Done in 14m 22s. Reproduction steps + evidence attached to every defect.
How it works

Six steps. One conversation. No human in the loop.

You hand your AI a URL and credentials. The MCP server gives the AI everything it needs to explore, write, save, run, and report.

1
Provide URL
App URL + credentials. That is it.
2
AI explores
Walks pages, classifies each, builds sitemap.
3
AI generates
Test cases for every meaningful flow.
4
Saved to SimplifyQA
TM, Object Repo, Suites, auto-linked.
5
AI executes
Via the existing SimplifyQA engine.
6
AI reports
Pass / fail / defects auto-filed in Jira.
The three pillars

Open. Universal. Zero-Touch.

Three decisions that make this product better as the AI ecosystem gets better.

01 · OPEN

Bring your own AI.

Claude Desktop, Cursor, Windsurf, ChatGPT, Anthropic SDK, OpenAI Assistants, LangChain, CrewAI. Anything that speaks MCP, or can be wrapped to. No proprietary client. No vendor lock-in.

02 · UNIVERSAL

All 9 technologies, day one.

Web, Mobile Web, API, Native Mobile (Android & iOS), Desktop, Database, Mainframe, SAP, Hybrid cross-system flows, plus a Generic type for anything else you throw at it. The AI picks the right test type.

03 · ZERO-TOUCH

URL + creds in. Tested app out.

No scripts. No new UI to learn. No flows to configure. No object repository to hand-build. Discovery, generation, save, execute, and reporting, the agent does it all.

What this replaces

Days of manual setup, condensed to 15 minutes in a chat.

Before

Regression suites, the old way

  • Hire QA engineers who know SimplifyQA
  • Click through the app manually
  • Author every test case by hand
  • Hand-build the object repository
  • Parameterize the test data
  • Assemble suites and execution plans
  • Schedule the runs
  • Days, sometimes weeks, per app
With Zero-Touch

One conversation in your AI

  • Open Claude / Cursor / Windsurf
  • Type: "Test https://… with user/pass"
  • AI explores 24 pages
  • AI generates 41 test cases
  • AI saves them in SimplifyQA
  • AI runs them across the right platform
  • AI files defects in Jira on failure
  • 15 minutes, end to end
What your AI can do

50+ MCP tools per technology, across the full testing lifecycle.

Every SimplifyQA capability exposed as an MCP tool, for every one of the 9 supported technologies. Your AI composes them in whatever order your prompt requires.

Exploration & Understanding
Walk the app, learn what it does.
  • Open a URL with credentials
  • Screenshot + DOM capture per page
  • Page-type classification
  • UI component detection
  • State recognition (loading, error, success, empty)
  • Application sitemap discovery
Test Case Generation
Platform-aware, RAG-grounded.
  • Functional, edge-case, negative, security, accessibility
  • Platform-correct tcType (WEB / API / ANDROID / IOS)
  • From BRDs, Jira, Azure, Figma, Excel, OpenAPI
  • Full feature → story → test hierarchy
  • Refine existing tests in place
Save to SimplifyQA
Native artifacts, not exports.
  • Features + stories with acceptance criteria
  • Test cases with steps + expected results
  • Object Repository auto-built
  • Parameterized test data iterations
  • Reusable suites assembled
  • Requirement ↔ test ↔ defect linkage
Execution & Evidence
Real runs, real evidence.
  • Trigger execution against any environment
  • Live progress stream (executing / passed / failed / healed)
  • Per-step screenshots + network + console + DOM + video
  • Self-healing for locator / timing / data / env drift
  • Auto-classify bug vs flake
Defects & Reporting
Triaged outcomes, not raw logs.
  • Auto-file defects to Jira with full evidence
  • Reproduction steps + severity classification
  • RTM on demand
  • Export PDF / Excel / CSV / TestLink XML
  • Push to existing CI/CD reporting
Integrations
Talk to the rest of your stack.
  • Jira / Azure DevOps, pull requirements, push defects
  • Figma, extract UI from wireframes
  • BRD / DOCX / XLSX / OpenAPI ingestion
  • Webhook triggers
  • Schedule recurring runs
All 9 technologies

Point your AI at any kind of application.

You point at…
The AI generates
What runs it
A web app
WEBtest cases
SimplifyQA web plugin
A mobile-responsive site
WEBwith viewport overrides
Web plugin
A REST or SOAP API
APItest cases
API plugin
An Android app
ANDROIDvia Appium
simplifyqa-android
An iOS app
IOSvia Appium
simplifyqa-ios
A Windows desktop app
DESKTOPvia SimplifyQADesktopDriver
desktop-execution-plugin
A database
DATABASEvalidation queries
simplifyqa-db-plugin
A mainframe terminal
MAINFRAMEscreen-flow tests
mainframe-plugin
An SAP transaction
SAPGUI tests
sap-plugin
A cross-system flow
HYBRID(Web → API → DB)
Multi-plugin orchestration
Anything else
GENERICtest cases (custom type)
Bring-your-own runner
Real conversations

What it actually looks like in your AI.

Each of these is a single prompt. Everything else happens automatically.

ACME Shop · QA
claude-opus-4.7
Test https://shop.acme.com with admin@acme.com / hunter2. Save to project ACME Shop. Run the suite.

I explored 24 pages and generated 41 test cases across login, browse, cart, checkout, and profile.

Saved as suite ACME_Checkout_v1 in your SimplifyQA Payments project.

Executed against staging: 37 passed, 3 failed (defects auto-filed in Jira: ACMESH-1041, 1042, 1043), 1 self-healed (locator drift on the promo banner).

Here's the RTM as Excel.

via SQA Zero-Touch Agent
Reply to Claude…
Compatible clients

If it speaks MCP, it works.

Claude Desktop
Paste server config into claude_desktop_config.json.
Cursor
MCP server registration. Tests without leaving the IDE.
Windsurf
MCP server registration. Engineers stay in flow.
ChatGPT
GPTs Actions or Assistants function calls.
Cline
VS Code extension. Register the MCP server and go.
Gemini
Google Gemini via function calling or the CLI.
opencode
Open-source AI coding CLI. Add the server to its MCP config.
VS Code
GitHub Copilot Chat / any MCP-aware VS Code extension.
Anthropic SDK
Any Python/TS app that speaks Anthropic messages.
LangChain / CrewAI
Register the server as a tool, orchestrate agents.
Setup in 60 seconds

Drop the server into your AI, done.

~/.config/claude/claude_desktop_config.json
{
  "mcpServers": {
    "sqa-agent": {
      "command": "npx",
      "args": ["-y", "@simplifyqa/zero-touch-agent"],
      "env": {
        "SQA_API_KEY":  "sk_live_…",
        "SQA_ORG_ID":   "acme"
      }
    }
  }
}

Your AI already knows what to do.

Give it a URL. It will handle the rest.

Back to SimplifyQA