Table of Contents

Great Question MCP

Great Question's MCP (Model Context Protocol) server gives your team read-only access to studies, sessions, transcripts, highlights, insights, and participant data from Claude Desktop, Claude Code, Cursor, ChatGPT, or any MCP-compatible AI tool.

What's available

V1 is in beta and provides read-only access to your research data:

  • Studies & sessions: list and retrieve study metadata and individual sessions
  • Full transcripts: complete interview transcripts for moderated and unmoderated sessions
  • Participant context: participant metadata scoped to your existing GQ permissions
  • Highlights & Insights: browse and search highlights, insights, and reels
  • Search: full-text search across studies, candidates, transcripts, sessions, and repository artifacts

Setup

All clients connect to the same server URL:

https://greatquestion.co/api/mcp/v1

You'll need an active Great Question account with MCP access enabled. The connection uses OAuth 2.1 — you'll authenticate through your browser with your existing GQ credentials. No API keys or secrets to manage.

Claude Desktop
  1. Open Settings → Connectors → Add custom connector.
  2. Enter the name Great Question and the server URL above.
  3. Click Save, then Connect. Authenticate via your browser.
  4. Restart Claude Desktop. GQ tools will appear when you start a new chat.
Claude Code

Run in your terminal:

claude mcp add --transport http GreatQuestion https://greatquestion.co/api/mcp/v1
ChatGPT
  1. Go to Settings → Apps → Advanced settings and toggle Developer mode on.
  2. Click Create app and enter:
    • Name: Great Question
    • Description: Query GQ Research data
    • MCP Server URL: https://greatquestion.co/api/mcp/v1
    • Authentication: OAuth
  3. Check "I understand and want to continue," click Create, and authenticate via your browser.
Cursor
  1. Open Cursor Settings (Cmd/Ctrl + Shift + J), then select MCP from the sidebar.
  2. Click + New MCP Server to open your mcp.json file.
  3. Add the following:
{
  "mcpServers": {
    "great-question": {
      "url": "https://greatquestion.co/api/mcp/v1"
    }
  }
}
  1. Save the file, then click Connect on the settings page. Authenticate via your browser.
Verifying your connection

Once connected, run through these quick checks in your MCP client:

  1. See your studies. Ask for your recent studies. You should see titles, statuses, and IDs you recognize.
  2. Find participants. Ask for candidates in one of those studies. Names and emails should appear as expected.
  3. Pull a transcript. Ask for transcripts from that study, then request the full content of one. Look for speaker labels and timestamps.
  4. Connect the dots. Pick a participant and ask which studies they've appeared in, what sessions they had, and what their transcripts say.
  5. Try a repo session. Ask for recent repo sessions, pick one, and request its transcript.

If something doesn't work, see Troubleshooting below.

Available tool calls

The MCP exposes the following tools. All tools are read-only.

Study tools

Tool

What it does

list_studies

Browse all studies in your workspace. Returns title, status, style, and configuration. Supports pagination.

search_studies

Find studies by title, status (draft/active/closed), duration, creation date, last activity date, or last updated date.

get_study

Get full details for a specific study by ID, including methodology and configuration.

Candidate tools

Tool

What it does

list_candidates

Browse all participants. Returns name, email, demographics, and participation history. Supports pagination.

search_candidates

Find participants by name/email, study ID, or updated-since timestamp.

get_candidate

Get a full candidate profile by ID.

Session & Transcript tools

Tool

What it does

list_repo_sessions

Browse all sessions (interviews, calls, observations). Returns title, summary, and timestamps. Supports pagination.

search_repo_sessions

Find sessions by title, candidate IDs, creator IDs, or study IDs.

get_repo_session

Get full session details by UUID, including tags and associated study/candidate.

get_repo_session_transcript

Get the complete speaker-attributed transcript for a session.

list_transcripts

Browse all transcripts. Returns metadata and content. Supports pagination.

search_transcripts

Full-text search within transcript content. Filter by candidate IDs or study IDs.

get_transcript

Get a complete transcript by ID with speaker attribution.

Artifact tools (Highlights, Insights & Reels)

Type

Tool

What it does

Highlights

list_repo_highlights

Browse all highlights (annotated excerpts from sessions). Returns text, title, and color. Supports pagination.

search_repo_highlights

Find highlights by title/text, candidate IDs, creator IDs, or study IDs.

get_repo_highlight

Get full highlight details by UUID, including description, tags, and session reference.

Insights

list_repo_insights

Browse all insights (research findings and stories). Returns title, description, and content type. Supports pagination.

search_repo_insights

Find insights by title/description, creator IDs, or study IDs.

get_repo_insight

Get full insight details by slug, including text, cover image, tags, and associated study.

Reels

list_repo_reels

Browse all reels (curated highlight video compilations). Returns title and subtitle. Supports pagination.

search_repo_reels

Find reels by title/subtitle, creator IDs, or study IDs.

get_repo_reel

Get full reel details by token, including clips and tags.

Common tool call parameters

Most list and search tools support these options:

  • Pagination: page (default 1), items (default 20, max 100), cursor for continued pagination
  • Sorting: orderBy — sort by createdAt or updatedAt (default)
  • Date filtering: Exact time windows or relative durations

Use cases

Prep for a stakeholder readout

Pull recent studies, surface key highlights and insights, and get transcript excerpts, all without leaving your AI tool. Ask something like: "Summarize the top findings from our onboarding study and pull the strongest supporting quotes from transcripts."

Catch up on research you missed

Search across studies and transcripts to quickly get up to speed. Try: "What studies have been completed in the last month? Give me a summary of each."

Find patterns across participants

Chain candidate lookups with transcript searches to spot themes. Example: "Which participants mentioned pricing concerns? What studies were they in, and what exactly did they say?"

Pull supporting evidence for a product decision

Search highlights and insights to build a case grounded in research. Ask: "Find all highlights related to navigation issues and summarize the common themes."

Audit participant coverage

Check which candidates have been contacted, how often, and across which studies. Try: "Show me all studies that participant Jane Doe has been involved in and their outcomes."

Troubleshooting

OAuth window doesn't appear: Make sure your browser allows popups from your MCP client. Some browsers block the OAuth redirect by default.

Tools not showing up: Restart your MCP client after saving the configuration. If tools still don't appear, confirm with your GQ account team that your user has the feature enabled.

Permission errors: The MCP respects your existing GQ access controls. If you can't see certain studies or sessions, check your permissions in the GQ product.

Usage Tips

  1. Start broad, then narrow: use list tools first, then search with specific filters.
  2. Chain queries: get a study ID, then find its candidates, then retrieve their transcripts.
  3. Use full-text search: most search tools support searching within titles, descriptions, and content.
  4. Leverage pagination: for large result sets, use pagination to process data in batches.
  5. Combine filters: most search tools accept multiple filters for precise results.

Security & data handling

Authentication

OAuth 2.1 with PKCE. No secrets stored on your device.

Permissions

Respects your existing GQ access controls. You only see what you can already see in the product.

Data scope

Read-only. Text-based data only — no video, audio, or biometric data is transferred.

Data freshness

Real-time. Queries hit the live API, not a cached or synced copy.

Retention

Response content is not stored or cached by default. Request metadata is retained per GQ's existing policies.

Third parties

The MCP server runs on GQ's infrastructure. No additional third-party processors are introduced. Once data is returned to your MCP client, it's handled by that client's data policies.

For full privacy details, see our Privacy Policy

Still need help?

Reach out to us anytime in the app or email us at [email protected]!

How did we do?

Contact