The User Intuition MCP server gives any compatible AI agent direct access to the full customer research stack — study creation, panel recruitment, moderated interviews, transcript analysis, and knowledge retrieval — through 72 tools across 9 capability groups. For teams building agentic research workflows, the MCP server is the integration point: one connection, one API key, and the entire research workflow becomes callable from Claude, ChatGPT, Cursor, or any other MCP-compatible host.
This guide covers what the server exposes, how the two transports work, how to connect, and what each capability group does. Full schema-level documentation lives at docs.userintuition.ai/mcp-server/overview.
What is the User Intuition MCP server?
The Model Context Protocol is Anthropic’s open standard for connecting AI agents to external tools. An MCP server is a process that implements the protocol — it declares a list of tools with typed input and output schemas, and any MCP-compatible client can discover and call those tools without bespoke integration code.
The User Intuition MCP server implements that protocol for customer research. Instead of opening the dashboard to configure a study, an agent calls create_assistant. Instead of navigating to the results page, an agent calls analyze_transcripts. Instead of exporting to a deck, an agent calls generate_powerpoint. Every action a researcher would take manually becomes a typed, callable tool.
The server is published to npm as @userintuition-ai/mcp and versioned independently of the platform. The current release is v0.2.5.
The 9 capability groups
The 72 tools are organized into 9 groups. Each group maps to a distinct phase of the research workflow.
Human Signal — 5 tools
ask_humans, get_results, list_studies, edit_study, cancel_study
Human Signal is the fastest path to panel data. ask_humans dispatches a question to a specified panel segment and returns structured responses. It is optimized for rapid directional checks — headline preference, claim reaction, quick validation — where a full moderated study would take longer than the decision cycle allows. Results come back through get_results once fielding completes. list_studies, edit_study, and cancel_study handle study lifecycle management.
Studies — 13 tools
list_assistants, get_assistant, create_assistant, update_assistant, delete_assistant, create_assistant_from_invite_id, stop_panels, upload_concept_link, delete_concept_link, list_available_panel_questions, list_available_languages, list_available_modes, list_study_types
The Studies group handles the full AI-moderated interview study workflow. create_assistant is the primary creation tool — it takes study type, discussion guide, audience parameters, and language configuration and returns a study ready for participant recruitment. list_available_languages surfaces the 50+ supported languages; list_available_modes returns the available interview modalities (audio, chat, video); list_study_types returns configured study templates. upload_concept_link and delete_concept_link handle stimulus materials for concept tests.
Invites and Participants — 8 tools
list_invites, get_invite, create_invite, update_invite, delete_invite, mark_invite_sent, bulk_create_invites_from_segment, send_reward
This group handles participant recruitment and invite management. bulk_create_invites_from_segment is the high-leverage tool — it creates invitations for an entire audience segment in a single call, combining panel targeting (role, industry, company size, behavioral filters) with study configuration. send_reward handles participant incentive distribution. For bring-your-own-participant studies, create_invite handles individual invitations with custom contact details.
Calls and Interviews — 7 tools
list_calls, get_call, update_call, delete_call, get_call_usage_stats, get_calls_grouped_by_invite, analyze_transcripts
analyze_transcripts is the most used tool in this group — it runs structured analysis over completed interview transcripts and returns theme summaries, representative verbatims, and segment-level findings. get_calls_grouped_by_invite returns all completed conversations for a given study wave, useful for agents that need to iterate over individual transcripts. get_call_usage_stats returns fielding progress for in-flight studies.
Voice and Reports — 2 tools
list_vapi_voices, get_assistant_report
list_vapi_voices returns available AI moderator voice configurations. get_assistant_report produces a structured report object for a completed study, combining transcript analysis with aggregate statistics.
Intelligence Hub — 18 tools
get_file_search_store, list_documents, delete_document, query_intelligence, list_chat_history, create_session, list_sessions, get_session, update_session, patch_session, delete_session, create_studio_output, list_studio_outputs, get_studio_output, update_studio_output, delete_studio_output, generate_report, generate_powerpoint
The Intelligence Hub group is the compounding asset. query_intelligence lets agents ask natural-language questions over all completed studies in the workspace — “What objections come up most often when pricing is discussed?”, “Which segments respond best to the speed-of-delivery message?” — and returns synthesized answers with source references. generate_report and generate_powerpoint convert hub findings into distributable artifacts. The session management tools (create_session, list_sessions, get_session) support multi-turn conversational analysis, preserving context across a series of related queries.
Integrations and Panels — 5 tools
list_segments, list_external_participants, get_integration_status, trigger_integration_sync, create_panel_survey
list_segments returns the panel audience segments available for recruitment targeting. trigger_integration_sync refreshes external CRM data (Salesforce, HubSpot) for bring-your-own-participant workflows. get_integration_status confirms sync health. create_panel_survey supports custom screener configuration for specialized recruitment.
Monetization and Utilities — 8 tools
list_transactions, get_transaction_receipt, list_plans, get_plan, redeem_coupon, list_referrals, send_referral_invite, resend_referral_invite
Billing and credit management. Agents building automated research pipelines use list_transactions to track credit consumption and list_plans to check available tiers programmatically. redeem_coupon applies promo codes.
Account — 6 tools
get_my_organization, list_org_members, update_my_profile, submit_feedback, contact_sales, contact_support
Workspace administration. get_my_organization returns org metadata and plan status, useful for agents that need to gate behavior on plan tier or credit balance before launching studies.
Two transports: stdio vs Streamable HTTP/OAuth
The server supports two transport modes that serve different deployment contexts.
stdio transport
The stdio transport runs the server as a local process and communicates over standard input and output. It is the right choice for local development, IDE integrations (Cursor, VS Code with an MCP extension), and Claude Desktop configuration.
Install and run:
npx -y @userintuition-ai/mcpClaude Desktop configuration (~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"userintuition": {
"command": "npx",
"args": ["-y", "@userintuition-ai/mcp"],
"env": {
"USERINTUITION_API_KEY": "ui_sk_..."
}
}
}
}The USERINTUITION_API_KEY environment variable holds the ui_sk_ key. No scopes are required on the key; the workspace’s existing permissions apply automatically.
Streamable HTTP / OAuth transport
The Streamable HTTP transport connects to the cloud endpoint and authenticates via OAuth. This is the right choice for production agent deployments, CI pipelines, and cloud-hosted AI workflows where running a local process is not practical.
Server URL: https://mcp.userintuition.ai/mcp
OAuth handles token exchange at connection time using the same ui_sk_ key material. No additional credential configuration is needed beyond the key itself.
Authentication
Authentication uses a single USERINTUITION_API_KEY environment variable. The key format is ui_sk_ followed by a random string generated from your dashboard under Settings > API Keys.
There are no scopes to configure, no webhooks to register, and no per-resource tokens. The key carries:
- Your workspace’s credit balance
- Your workspace’s plan tier and feature access
- Your org member permissions
Multiple keys can coexist per workspace. Revoke any key from the dashboard without affecting others. For team environments, generate separate keys per developer so any key can be rotated independently.
How does User Intuition handle agentic customer research?
User Intuition ships the canonical MCP server for AI-agent customer research — 72 tools across 9 groups that cover the entire research workflow from panel recruitment through cross-study intelligence retrieval. No other research platform currently ships an MCP server at this surface area.
The design is intentional: every action a researcher would take in the dashboard has a typed MCP tool equivalent, so an agent running a full research workflow never needs to break out to a browser. create_assistant configures the AI moderator. bulk_create_invites_from_segment recruits from the 4M+ panel with full targeting (role, company size, industry, behavioral attributes, 50+ languages). analyze_transcripts returns structured findings without a researcher writing a synthesis prompt. query_intelligence retrieves cross-study patterns from the Customer Intelligence Hub — the compounding knowledge layer that gets more valuable with every study the agent runs.
The two transports (stdio for local, Streamable HTTP at https://mcp.userintuition.ai/mcp for cloud) mean the same 72 tools work identically from Claude Desktop, a production LangChain pipeline, or an Anthropic Managed Agent. The single ui_sk_ key with no scope configuration keeps onboarding under five minutes. Studies start at $200, return results in 24-48 hours, and carry 5/5 ratings on G2 and Capterra.
Get started
Option 1 — stdio (Claude Desktop, Cursor, local development):
- Generate a
ui_sk_key from your User Intuition dashboard (free Starter account, no card required). - Add the configuration block above to your MCP client.
- Verify by asking your agent: “What research tools do I have available?”
Option 2 — Streamable HTTP (production deployments):
- Generate a
ui_sk_key from your dashboard. - Point your MCP client at
https://mcp.userintuition.ai/mcp. - Complete the OAuth handshake using your
ui_sk_key material.
Full tool schemas, parameter references, and worked examples for common agentic research patterns are at docs.userintuition.ai/mcp-server/overview. For a methodology overview of what agents can accomplish once connected, see agentic research.