1. Docs
  2. AI
  3. MCP & AI assistants
  4. FAQ & limits

FAQ & limits

Answers to common questions about AI assistants in SmartestQA and the limits of the MCP server.

3 min read Updated CustomersAdmins

In short

On this page you can

  • Get short answers about clients, privacy, access and revoking it.
  • Filter the questions for customers or admins.
  • Run a five-minute MCP demo for a prospect.
  • Look up every MCP limit in one table.

How to use it

  1. Pick your audience in the filter bar above the questions.
  2. Open a question; follow its link for the full procedure.
  3. Check Limits when the assistant stops waiting or a preview expires.

Questions and answers#

Which assistants can I use?Customers

Claude on claude.ai (through the connector), Claude Desktop (extension or command), Claude Code and Cursor (command), and any other MCP client that can start a local server. See Choose a way.

Connector or extension: which should I take?Customers

On claude.ai, the connector: no install, you sign in with your SmartestQA account. On your own computer with Claude Desktop, either works; the extension uses your host’s API key. For Claude Code or Cursor, the command.

Does SmartestQA see my conversations?Customers

No. SmartestQA receives only the tool calls the assistant makes, with their arguments, and answers them. The conversation itself stays with your AI provider.

Which SmartestQA data goes to the AI provider?Admins

The results of the tools the assistant calls: for example project status, requirement names, steps, run logs, defects. They go to the provider of the assistant (for Claude, Anthropic) under that provider’s terms. Use test environments and test data. Passwords are never sent: they are data source keys you fill in SmartestQA.

Can a user reach more through MCP than in the app?Admins

No. Every call runs as that user, with the same project access and permissions. Removing records also needs an Admin role and a click in a dialog. See Other safeguards.

Can I give someone read-only access through MCP?Admins

For a local install, yes: set SMARTEST_READONLY=true in the server’s environment (see Read-only connection); the assistant then sees only reading tools. The connector has no read-only switch; its access follows the user’s role.

How do I take access away?Admins

Connector: disconnect SmartestQA in Claude under Settings › Connectors; a token already issued expires within the hour. Local MCP: its key stops working when its host is deleted in Test Agents; also remove the extension or run smartest-mcp --uninstall.

Can I see what an assistant changed?Admins

Yes. ConfigurationLogsAudit logs records every change with who made it and from where; MCP changes show the source MCP (Claude).

Does it work for mobile and desktop projects?Customers

Yes. The step rules cover web, mobile and desktop steps. In a MOBILE project with Android and iOS, the assistant asks which platform to dry-run on. Desktop and local mobile runs need your agent online.

Where do the assistant’s dry runs run?Customers

On the SmartestQA server or on your online agent (see where dry runs run); the assistant asks or follows what you say (“on my agent”). See Do you need an agent?.

How is this different from Autonomous Agents?Customers

With MCP you work with the assistant in a conversation and approve each change as you go. Autonomous agents run alone on a schedule or after a deploy and leave a report and proposals for later.

Do I need a particular Claude plan?Customers

Claude decides which of its plans allow custom connectors and extensions; check your plan in Claude. In a Claude Team or Enterprise organization an owner may have to allow the connector first. On the SmartestQA side, your company’s licence decides what you can use; an Admin sees it in Company Settings › Billing.

Limits#

ItemLimit
Tools · ready-made prompts18 · 4
A preview stays valid30 minutes, once, while the MCP keeps running
Red dry runs before a scenario is BLOCKED3
Wait for a dry run result per call45 s by default, at most 50 s (then the assistant asks again)
Wait for a pipeline result per callat most 50 s
Connector access token · renewal1 hour · 30 days
Connector sign-in attempts10 failures in 15 minutes, then a 15-minute pause
Node.js for the command and the zip18 or newer
Claude Desktop extensionWindows and macOS
Tool resultsin English; the assistant answers in your language

Didn’t find your answer?Write to support with your project name and what you see on screen.

info@smartestqa.ai
© 2026 Selenity Technology · SmartestQAsmartestqa.aiThis page as MarkdownChecked against smartest-mcp-server 1.31.1, smarttest-fe predev 2a8c79cc, smarttest-be 633c2a03 (oauth, ApiKeyAuthenticationFilter)