Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallThe Applitools MCP Server lets an AI assistant help configure and author visual tests, investigate existing Eyes results, and resolve baseline changes. The key limitation: the documented setup and checkpoint-authoring tools support Playwright TypeScript/JavaScript Fixtures, while result inspection, review, and resolution can work with existing Eyes results produced by other supported SDKs.
What the MCP server can—and cannot—maintain
Applitools describes its MCP server as helping developers create, update, review, and resolve visual tests in Eyes. Eyes tests capture screenshots and compare them with stored baselines: the expected appearance for a particular test state and environment. A difference is a review item, not automatically a defect or a safe baseline update. Applitools Eyes overview
| # | Preview | Product | Price | |
|---|---|---|---|---|
| 1 |
|
AI FOR QUALITY ASSURANCE AND SOFTWARE TESTING: The Practitioner's Complete Guide to AI-Powered... | $9.99 | Buy on Amazon |
| Work | Documented scope |
|---|---|
| Project setup, UFG configuration, and adding checkpoints | Playwright TypeScript/JavaScript Fixtures; tools need access to project source code. |
| Inspecting and reviewing captured results | Existing Eyes batch, session, step, and DOM data from supported SDKs and languages. |
| Resolving checkpoints and match regions | Existing Eyes results; changes that save or reset baselines require a separate explicit request and approval. |
That distinction matters if your team uses another SDK: the server may still help investigate its Eyes results, but the documented setup and checkpoint-insertion tools do not thereby become available for that framework.
Check requirements and credentials first
Runtime and MCP client
- Use Node.js 18 or newer.
- Connect through an MCP-capable assistant/client. Applitools recommends its VS Code or Cursor extensions; manual setup is also documented for clients that accept a stdio server configuration.
- For setup and checkpoint authoring, use a project based on Playwright TypeScript/JavaScript Fixtures and allow the assistant to access the source code.
Client extension menus and configuration formats can change. Follow Applitools’ current MCP Server documentation for the exact steps and current supported-client instructions.
#1 Best Overall
Use only the key appropriate to the task
| Credential | Use |
|---|---|
APPLITOOLS_API_KEY |
Test execution. |
APPLITOOLS_READ_KEY |
Inspection and review. |
APPLITOOLS_WRITE_KEY |
Resolution tools and review in resolve mode. |
Inspection requires a read key. Review always requires a read key; resolve-mode review also requires a write key. Resolution tools require a write key. Keep write access out of a read-only investigation workflow where your credential controls allow it. The documentation says missing credentials produce an error that identifies the missing key, and key values are not exposed in logs, errors, or tool responses.
Install and connect the MCP server
Applitools recommends installing through its VS Code or Cursor extensions. For a manual client configuration, the documented stdio server command is:
npx --yes @applitools/mcp@latestConfigure that command in the MCP client’s server settings, then provide the task-appropriate credentials through the client’s documented environment-variable mechanism. Exact UI paths differ by client, so use the current client-specific instructions linked from Applitools’ MCP documentation rather than copying a configuration format intended for another assistant.
Configure a Playwright Fixtures project and add checkpoints
For a Playwright TypeScript/JavaScript Fixtures project, the documented tools include eyes_setup_project, eyes_setup_ufg, and eyes_add_checkpoints_to_test. They can help configure the project and insert visual checkpoints, but they are not a general-purpose authoring path for every Eyes SDK.
- Open the project in the connected MCP client and confirm the assistant can access the source files.
- Ask it to configure the Eyes project, or configure Ultrafast Grid (UFG), as needed. Review the proposed edits against your existing Playwright setup before accepting them.
- Ask it to add checkpoints to the relevant test or state. Inspect the resulting code and ensure each checkpoint represents a stable, meaningful UI state.
- Run the test using the project’s normal test command and the appropriate execution credentials. The Eyes SDK captures screenshots and sends them to Eyes for comparison with stored baselines.
Use the current tool documentation for the supported arguments and recommended SDK versions; the tool names and scope are documented, but this article does not assume a universal Playwright test command or a fixed generated code shape.
Investigate a visual difference before changing a baseline
Start with a read-only question, scoped to the affected batch, scenario, session, or step. The review workflow can gather evidence such as images, DOM differences, and history. Read-only tools can also expose sessions and batch statistics, search DOM data, show active match regions, and trace a node across runs.
“Review my last batch and tell me what changed”
Ask the assistant to inspect the relevant batch and summarize the differences. Confirm that it selected the intended batch, then use the returned images and available DOM/history evidence to locate the changed state. A batch-level view is useful for triage; drill into the affected scenario, session, or step before deciding whether the change is expected.
“Just show me what changed in this batch, don’t resolve anything yet”
State explicitly that the request is read-only and that no checkpoint or baseline should be changed. Keep the operation in inspect mode. This separates evidence gathering from a later decision that could affect what future runs treat as expected.
“Is this diff on the timestamp element dynamic, or a real change?”
Compare the changed image area with DOM differences and the node’s history across runs, if those data are available. A timestamp that varies across runs may be dynamic; a single visual diff alone does not establish that it is harmless. Check whether the variation is expected for the test and whether a suitable match region is already active before proposing a change.
DOM-based inspection depends on DOM capture having been enabled when the test ran. If it was not enabled, session or DOM tools may return no data; use the available screenshot and other result evidence rather than treating the missing DOM as proof that nothing changed.
Resolve checkpoints and match regions deliberately
After reviewing the evidence, ask for a specific resolution only if the change is understood. Resolution tools can accept or reject checkpoints and add, remove, or update match regions. Review can run in inspect or resolve mode, but review itself does not save an accepted or rejected result to the baseline.
Applitools documents the boundary explicitly: “Saving changes to the baseline, or resetting it, is never part of Review itself; ask separately for eyes_resolve_save (or eyes_resolve_reset), each of which still requires your explicit approval.” Treat an assistant’s explanation as a recommendation. Inspect the proposed checkpoint or region action, and give the required approval before saving or resetting a baseline.
Choose the workflow that fits your team
- You need new setup or checkpoint code: confirm the project uses Playwright TypeScript/JavaScript Fixtures and that source access is available.
- You only need to understand existing failures: use inspection and review on the existing Eyes results; these tools are not limited to the Playwright authoring scope.
- You rely on DOM evidence: verify that DOM capture was enabled in the run being investigated.
- You need controlled approvals: separate read-key investigation from write-enabled resolution where practical, and require explicit approval for saves or resets.
- Your client cannot run MCP stdio servers: follow the client’s current integration instructions or use a workflow that your client supports; the documented manual example is a stdio command.
Troubleshooting
Server does not start
Check that Node.js 18 or newer is installed and that the MCP client is configured to launch npx --yes @applitools/mcp@latest as a stdio server. Client setup differs, so compare the server entry with the current instructions for that client.
A tool reports a missing key
Provide the credential for the operation: API key for test execution, read key for inspection and review, and write key for resolution or resolve-mode review. Do not substitute the execution key for the read/write permissions.
Setup or checkpoint tools are unavailable
Check that the project is Playwright TypeScript/JavaScript Fixtures and that the assistant can access its source code. The documented setup and checkpoint tools have this narrower framework scope.
Session or DOM inspection returns no data
DOM-based inspection needs DOM capture from the original run. If capture was off, those tools may have nothing to return; investigate with the image and other available Eyes result information.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsThe assistant proposes a baseline update during review
Do not treat the proposal as a committed change. Review findings and baseline writes are separate; request the appropriate save or reset action only after deciding what should become the expected result, and approve it explicitly.
Or skip the browser setup
If you need a screenshot of a page without configuring a browser capture script, ScreenshotNeo provides a website screenshot API and MCP server. One GET request returns an image or PDF; this example saves a WebP screenshot:
Quick Recap
curl -G "https://api.screenshotneo.com/v1/shot" -d access_key=YOUR_API_KEY --data-urlencode url=https://stripe.com -o shot.webpSee the ScreenshotNeo API documentation for request options. Cookie banners, newsletter popups, and chat widgets are removed before capture; bot checks, blank pages, and failed loads are never billed. Its MCP server lets AI agents take screenshots, and the free plan includes 1,000 screenshots a month with no card; paid plans start at $5 for 3,000. Sign up for free ScreenshotNeo access.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
