MCP server
Get your user research done in
Claude Code
On every plan including Free. No experience required.
Claude Code
Cursor
Codex
Lovable
v0
Replit
Bolt
https://mcp.fiuto.aiAdd it as an MCP server in your agent.
What it does
The same loop as the browser, driven from the agent. 32 tools.
- 1
Create a study, plan-first
create_study, save_plan, build_from_plan: the agent drafts a plan you review before anything is built, the same review step the web app uses.
- 2
Launch, and wrap a prototype URL
launch_study plus wrap_url and attach_tracker for a live prototype or site the study needs to open.
- 3
Read the results back
get_study_results and export_study_results bring responses, findings, and traces back into the agent session.
What reads and writes
Same visibility rules as the web app, on every plan including Free.
Why this is different
Fiuto's MCP creates, launches, and reads back. Launch is the verb none of the others have. What the others expose, in their own words.
Lyssna
Read-only, in beta, Claude and ChatGPT, on paid plans.
Creates, launches, and reads back. Seven agents. Every plan including Free.
Maze
Brings findings into Claude, ChatGPT, Gemini, Copilot, and Cursor. Described as reading insights, not creating studies.
Creates, launches, and reads back. Seven agents. Every plan including Free.
UserTesting
Read and write: recruit, create studies, launch tests. Claude, ChatGPT, Figma Make. Limited early access for select teams.
Creates, launches, and reads back. Seven agents. Every plan including Free.
Integrations
Nothing else is listed.
Prolific recruitment
Order real respondents from Prolific's global panel of 300,000+ participants, with demographic filters, from the launch page.
Export API
Raw results as JSON or CSV through a per-study token you can rotate, on every plan including Free.
Frequently asked questions
01Is Fiuto's MCP server free?
Yes, on every plan including Free. The MCP server follows the same visibility rules as the web app, so what it can see and do matches the account it belongs to, not a plan floor.
02Do I need a paid plan to use it?
No. Every tool works on Free. What runs against your AI credits (drafting a plan, an AI tester) costs the same over MCP as it does in the browser. The plan only changes how many credits and live studies you have.
03Can the agent launch a study without me?
It builds the plan first, block by block, for you to review, the same review step the web app uses before anything goes live. From there, it can launch the study for you if you ask it to.
04What about my data?
Yours. The same per-study API token, JSON/CSV export, and account-level access rules apply over MCP as in the browser, on every plan including Free.
05Which model does the agent use?
None. That's the point. Fiuto's MCP server doesn't run its own model; it reads and writes through the coding agent you already have open, whichever model that agent is running.
06Does Fiuto work with Cursor?
Yes. Add https://mcp.fiuto.ai as an MCP server in Cursor and sign in when it asks. Every one of the 32 tools is available: create a study, launch it, read the results back.
07Does Fiuto work with Lovable?
Yes. Add https://mcp.fiuto.ai in Lovable's MCP settings. The same tools are available as in any other client, and Fiuto's tracker can wrap the prototype you are building there.
08Does Fiuto work with v0?
Yes. Add https://mcp.fiuto.ai as an MCP server in v0 and sign in when prompted. Nothing about the tool list changes between agents.
09Does Fiuto work with Replit?
Yes. Add https://mcp.fiuto.ai in Replit's MCP settings. You can wrap a Replit-hosted prototype and read sessions back into the same chat.
10Does Fiuto work with Bolt?
Yes. Add https://mcp.fiuto.ai as an MCP server in Bolt and sign in when it asks. The full 32-tool surface is available on every plan including Free.
Connect your agent, or start in the browser.
Add the endpoint to the agent you already build in, or sign up and build the study by hand. Either way you review every block before it launches.
Connect your agent