Interview MCP Server

Model Context Protocol server providing 10 tools for AI-guided human interviews at scale. Create structured questionnaires, distribute share links, monitor real-time participation, and generate aggregated statistics — all through MCP.

Server Endpoint

https://atypica.ai/mcp/interview

Authentication: Bearer token with your API key (atypica_xxx)

Available Tools

Tool	Description
atypica_interview_ create_project	Create an interview project with a brief and structured questions
atypica_interview_ update_questions	Replace the question list (doesn't affect started sessions)
atypica_interview_ list_projects	List and search your interview projects with pagination
atypica_interview_ generate_share_link	Generate a permanent or temporary share URL for interviewees
atypica_interview_ get_progress	Get participation breakdown (human vs. persona, completed vs. in progress)
atypica_interview_ list_sessions	List sessions for a project with status filtering and pagination
atypica_interview_ get_session	Get full session detail including structured answers and transcript
atypica_interview_ get_stats	Read aggregated question statistics (ready / generating / not yet triggered)
atypica_interview_ generate_stats	Trigger async generation of aggregated statistics (30-90s)
atypica_get_ upload_credentials	Get a presigned upload URL for question images

Quick Start

1. Get your API key from account settings

2. Connect to the MCP server with authentication:

curl -X POST https://atypica.ai/mcp/interview \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "method": "tools/call",
    "params": {
      "name": "atypica_interview_create_project",
      "arguments": {
        "brief": "Understand coffee brand preferences",
        "questions": [{ "text": "What brand do you usually buy?" }]
      }
    },
    "id": 1
  }'

Interview Workflow

Unlike Study MCP, human interviews are fully asynchronous — real people participate on their own schedule, from minutes to days later.

Create an interview project with a brief and structured questions
Distribute the generated share link to interviewees
Monitor progress — poll every 5-10 minutes as responses come in
Generate aggregated statistics once enough sessions complete
Retrieve aggregated stats, or drill into individual session transcripts and structured answers

Question Types

Type	Options	Notes
open	Must not have options	Text answer; AI probes for detail
single-choice	Must have ≥2 options	One selection only
multiple-choice	Must have ≥2 options	Multiple selections allowed

Questions support up to 4 image attachments each (10MB max per image), uploaded first via atypica_get_upload_credentials.

atypica-interview Skill

The atypica-interview skill wraps the tools above with usage guidance for MCP-compatible AI assistants — including how to monitor asynchronous progress and when to trigger statistics generation.

Download

atypica-ai/atypica-interview-skill

Direct download • Source code • Issues & contributions

Installation Steps

Get API Key: Visit https://atypica.ai/account/api-keys to create an API key (format: atypica_xxx)

Install MCP Server: Add the server to your AI assistant's MCP configuration

# Example command (syntax varies by tool)
mcp add --transport http atypica-interview https://atypica.ai/mcp/interview \
  --header "Authorization: Bearer YOUR_API_KEY_HERE"

Restart: Restart your AI assistant to load the new tools

API Reference

The skill package's reference file is the single source of truth for detailed input/output schemas, completion semantics, error codes, and security notes for all 10 tools — kept in sync with the server on every release:

atypica-interview-skill/references/api-reference.md

Support & Community

API Keys: Manage your API keys in account settings

Questions? Contact us through our support channels

atypica.AI Docs