
sunzu
com.sunzu/sunzu
User research with synthetic humans: interviews, concept, usability and ad tests in minutes.
Documentation
Sunzu MCP server
Run user research from your AI tool. Sunzu's remote MCP server lets Claude, ChatGPT, Cursor, VS Code and other MCP clients run depth interviews, concept tests, usability tests and content (ad) tests with synthetic humans grounded in customer evidence, then read the report in chat.
- Server URL:
https://app.sunzu.com/mcp(Streamable HTTP, OAuth sign-in) - Registry name:
com.sunzu/sunzu - Setup guide: https://sunzu.com/mcp/
- Demo: https://sunzu.com/videos/mcp/
- Pricing: https://sunzu.com/pricing/ (Solo's first month is free)
No account yet? Approving the connection in your browser lets you create one, then brings you straight back.
Findings are simulated: directional signal from research-grounded synthetic humans. They complement research with real people.
Install
Claude (web, Desktop, Claude Code with the same account): add Sunzu as a custom connector, then click Connect.
Claude Code
claude mcp add --transport http --scope user sunzu https://app.sunzu.com/mcp
claude mcp login sunzu
ChatGPT: open chatgpt.com/plugins, click +, then Add custom MCP server. Name it Sunzu, paste https://app.sunzu.com/mcp, choose OAuth, then Create as a plugin.
Cursor: add Sunzu to Cursor
VS Code (GitHub Copilot): add Sunzu to VS Code
Any other client, as JSON:
{ "mcpServers": { "sunzu": { "url": "https://app.sunzu.com/mcp" } } }
Terminal and coding agents
npx @sunzuai/cli login
npx @sunzuai/cli --help
Tools
| Tool | What it does |
|---|---|
get_workspace, get_credit | The workspace this connection uses and the credits left |
list_audiences, add_library_audience | Audiences you can study, and free library audiences to add |
quote_study | The price of a study, in credits, before it starts |
prepare_upload_page, prepare_stimulus_upload, get_uploaded_files | Upload concepts, ads or prototypes |
start_depth_interview, plan_depth_interview, update_study_plan | Depth interviews, with a plan you can review first |
start_concept_test | Reactions to 1 to 3 concepts |
start_ad_test | How 1 to 5 versions of an ad or content land |
start_usability_test | Synthetic humans use a website, prototype, Figma flow or local app |
launch_study, resume_study | Launch a study created for review, or resume one |
wait_for_study, get_study_status, list_studies, get_report | Follow studies and read reports |
Launching a study spends your plan's credits. The connector never buys plans or credits: those stay in the app.
Privacy and security
- Privacy policy: https://sunzu.com/legal/privacy/
- Data processing agreement: https://sunzu.com/legal/data-processing/
- Disconnect any AI tool from your account page in the Sunzu app.
- Security contact: hello@sunzu.com
Publishing
server.json is published to the official MCP Registry by .github/workflows/publish-mcp-registry.yml on every v* tag. The namespace is proved by https://sunzu.com/.well-known/mcp-registry-auth; the matching private key is the MCP_REGISTRY_PRIVATE_KEY repository secret.
This repository holds metadata and docs only. Sunzu's product code is not open source.
© 2026 Sunzu, Inc.
No installation packages available.