Agent templatev1.0Updated 2026-11-03

BrowserStack Device Lab

Send a website QA brief to a dedicated device-testing service. Choose real devices, then review the service evidence and reproducible findings.

  • Physical devices through your own BrowserStack connection
  • Evidence, reproduction steps and severity in one report
  • Connection required before device tests can run
Clone this project Use from Claude / Codex

By FlyMy.AI templates/website-qa-browserstack v1.0

BrowserStack Device Lab: BrowserStack Device Lab with BrowserStack
BrowserStack Device Labgpt-6-luna
BrowserStack

$ YOUR NEXT RUN

Claude CodeCodexCursorAny agent

Use BrowserStack Device Lab in your assistant

Paste this prompt into Claude Code, Codex, Cursor or any agent. It reads the guide and helps you connect.

Read https://flymy.ai/templates/website-qa-browserstack/template.txt and help me set up BrowserStack Device Lab using my own connections. Ask for the missing inputs and budget before running it.
Open public projectDownload brief

Proof

  • Public projectCopy it to your account
  • Copy checkedSetup uses your own inputs and connections
  • Connection requiredNo real-device run verified yet
1agentone QA agent
1toolspriced below
on demandrunsschedule after the first report
Usage-basedconfirm budget firstyour usage determines the cost

What it does

  • Physical devices through your own authorized BrowserStack connection. Every report must identify the device, OS, browser, session and evidence returned by the service.
  • Check text fit, horizontal overflow, missing images and sticky overlaps.
  • Open and close menus, then follow navigation within the selected routes.
  • Report evidence, reproduction steps and severity; list checks that could not run.

Ideal for Teams checking public websites before a release, with explicit coverage and reproducible findings.

Connects to

  • BrowserStackYour own planyour account

Project diagram

from the reference project
Project diagram of BrowserStack Device LabINPUTS AND TIMERSLEADTOOLSOUTPUTSPublic website URLRoutes and QAfocusDevicesBrowserStackDevice Labgpt-6-lunaBrowserStackYour own planCoverage reportReproduciblefindingsBUDGET RULE / Up to 4 devices, one session per device, 5 minutes per session; confirm your plan and agent budget first.
  1. Inputs
    Public website URLRoutes and QA focusDevices
  2. Lead agent
    BrowserStack Device Labgpt-6-luna
  3. Tools
    BrowserStackYour BrowserStack plan; billed separately from FlyMy.AI agent usage.
  4. Outputs
    Coverage reportReproducible findings
The diagram as text
  • Inputs: Public website URL, Routes and QA focus, Devices
  • Lead: BrowserStack Device Lab (gpt-6-luna) uses BrowserStack
  • Tool: BrowserStack - Your BrowserStack plan; billed separately from FlyMy.AI agent usage. (your own account)
  • Output: Coverage report (from BrowserStack Device Lab)
  • Output: Reproducible findings (from BrowserStack Device Lab)

Day one

from the reference project
  • 01Copy projectOpen the public project and copy it to your account.no template fee
  • 02Connect toolsCreate or use a BrowserStack account with access to the device-testing product you need. Add the official remote MCP server in FlyMy.AI, select OAuth and sign in. Never paste account secrets into an agent chat.your plan
  • 03Set the scopeChoose the public website, up to 8 routes, profiles or devices, QA focus and budget.before running
  • 04Review the reportReview completed checks, findings and any partial or unavailable coverage.usage-based

How it runs

templates/website-qa-browserstack v1.0

Start after your inputs and budget are confirmed. BrowserStack must be authorized before a device session can begin.

The agent team

from the reference project

BrowserStack Device LabQAlead

Send a website QA brief to a dedicated device-testing service. Choose real devices, then review the service evidence and reproducible findings.

BrowserStack
Instruction excerpt

Physical devices through your own authorized BrowserStack connection. Every report must identify the device, OS, browser, session and evidence returned by the service. Stay within the selected routes and budget. Report only checks supported by returned evidence; mark incomplete coverage. Do not log in, submit forms, buy or send messages.

Budget: Up to 4 devices, one session per device, 5 minutes per session; confirm your plan and agent budget first.

on demandgpt-6-luna

You will be asked

once, in one message
  • Which website?A public HTTPS website; pages behind login are outside the scope.https://example.com
  • Which routes?Choose up to 8 routes within the website./, /pricing, /contact
  • Which devices?Up to 4 devices supported by your BrowserStack product.iPhone and Android
  • What should the report focus on?Choose the pages or interactions that matter for this check.menu, text fit, missing images
  • What budget should it use?Up to 4 devices, one session per device, 5 minutes per session; confirm your plan and agent budget first.Confirm before the first run

Keys and passwords never go into the chat: you connect accounts on their own pages.

What it costs

Run limits checked on 2026-11-03

0%template markup or purchase feeUsage pricing →

Price of one run
ItemUnit pricePer runCost
BrowserStackYour BrowserStack planUp to 4 devices x 5 minutesYour plan
FlyMy.AI agentModel usageActual work performedUsage-based
TemplateCopy the public projectOne copy$0
Final run costUsage-based

Your BrowserStack plan and FlyMy.AI agent usage are separate. At most 4 devices, one session per device, 5 minutes per session. The project cannot purchase a plan or extra capacity. No fixed-price result or device run is claimed.

What stays yours

the same for every template

Your connections stay yours

A copy never carries the author's keys or connections.

Nothing runs until you answer

Clone creates the agents and a setup chat; nothing bills before your reply.

A budget you set

Choose the budget before the first run. Review usage in your own project.

Versions

a copy keeps the version it was made from
  • v1.02026-11-03Public project with checked copy and connection-required flows; real-device tests await an authorized account.

Other ways to start

01 - Clone in FlyMy.AI

Open the reference project and press Clone; a setup chat asks once.

Clone this project
02 - Your Claude or Codex

Connect FlyMy.AI with skill.md and paste the request below.

03 - API and SDK

Agents MCP projects_copy with fleet:c625ca88-5da6-40df-88bf-a138efeddbd2, or client.projects.copy.

FAQ

Is this ready to run?

The copy and connection-required flow are checked. Real-device execution is ready for setup and has not been tested because no BrowserStack account is connected.

How do I connect BrowserStack?

Create or use your BrowserStack account with the device-testing product you need. In FlyMy.AI MCP connections, add https://mcp.browserstack.com/mcp with OAuth and sign in through BrowserStack. Never paste secrets into the agent chat.

What if OAuth registration is rejected?

The current connection received oauth_registration_failed. Creating an account alone does not confirm that this integration issue is resolved. A provider-supported OAuth client or platform connector is required. The callback is https://backend.flymy.ai/api/v1/agents/mcp-servers/oauth/callback/. The project stays at CONNECTION_REQUIRED until authorized.

Can I use the local BrowserStack MCP?

BrowserStack documents a local MCP with account credentials, but that does not automatically connect a cloud FlyMy.AI agent. Use a supported cloud connection for this project.

Does it submit forms or make purchases?

No. It checks public pages and navigation within your chosen routes. Forms, logins, messages and purchases are outside its scope.

What does it cost?

Your BrowserStack plan and FlyMy.AI agent usage are separate. At most 4 devices, one session per device, 5 minutes per session. The project cannot purchase a plan or extra capacity.

Does copying enable a schedule?

No. Copying starts with your own inputs and accounts. No schedule is enabled by default.

Illustrative example, not a live result
Illustrative report structure
website: https://example.com
coverage: complete, partial or connection required
profile: requested and observed dimensions, or device / OS / browser
finding: page, severity, reproduction, expected and observed result
evidence: service-returned details
unverified: checks that did not run

Sources

Public reference project →Usage pricing →MCP connections →BrowserStack remote MCP →BrowserStack local MCP →

Content and connection guidance checked 2026-11-03. See linked documentation for current availability and prices.