Claude Code Integration

Use Tabular Pro as a skill in Claude Code with SKILL.md for natural language survey workflows.

Claude Code Integration

Tabular Pro integrates natively with Claude Code through two mechanisms: the MCP Server (tool-level integration) and SKILL.md (high-level skill definitions). Together, they let you manage surveys, analyze data, and build dashboards using natural language.

Quick Setup

1. Generate an API Key

Go to Settings > API Keys in Tabular Pro and create a key.

2. Configure the MCP Server

Add to your project's .claude/settings.json:

{
  "mcpServers": {
    "tabularpro": {
      "url": "https://tabularpro.com/api/mcp",
      "headers": {
        "Authorization": "Bearer tp_your_key_here"
      }
    }
  }
}

3. Add a SKILL.md File (Optional)

Create a SKILL.md in your project root to teach Claude Code how to use Tabular Pro effectively. This gives Claude high-level context about your research workflows.

SKILL.md Template

Create a SKILL.md file in your project:

# Tabular Pro Survey Platform

## Overview
Tabular Pro is our market research platform. Use the tabularpro MCP server
tools to interact with surveys, responses, dashboards, and AI analysis.

## Common Workflows

### Analyzing Survey Data
1. Use `list_surveys` to find the target survey
2. Use `get_response_stats` to check response counts
3. Use `get_summary_statistics` for quantitative overview
4. Use `generate_insights` for AI-powered findings
5. For open-ended questions, use `run_sentiment_analysis` and
   `run_theme_extraction`

### Creating a Survey
1. Use `create_survey` with a descriptive name
2. Use `create_page` to add pages
3. Use `create_question` to add questions with appropriate types
4. Use `publish_survey` when ready for data collection

### Building a Dashboard
1. Use `list_surveys` to identify data sources
2. Use `create_dashboard` with a meaningful name
3. Use `get_chart_data` to preview visualizations
4. Share via the dashboard sharing endpoints

### Generating Reports
1. Use `get_response_stats` and `get_summary_statistics` for data
2. Use `generate_insights` for AI analysis
3. Use `export_report_pdf` to generate deliverables

## Question Types
- `single_choice` — Radio buttons (one answer)
- `multiple_choice` — Checkboxes (multiple answers)
- `text` — Open-ended text
- `number` — Numeric input
- `scale` — Linear scale (e.g., 1-10)
- `nps` — Net Promoter Score
- `matrix` — Grid/matrix
- `ranking` — Drag-and-drop ranking

## Conventions
- Always check `workspace_id` context when listing data
- Survey IDs are required for most response/analysis operations
- AI operations consume credits — confirm with user before bulk operations
- Export formats: csv, xlsx for data; pdf, pptx for reports

How It Works

When Claude Code has both the MCP server configured and a SKILL.md file:

  1. Discovery: Claude sees the available Tabular Pro tools via MCP
  2. Context: SKILL.md provides workflow knowledge and conventions
  3. Execution: Claude calls MCP tools to fulfill your requests
  4. Intelligence: Claude combines tool results with its reasoning

Example Conversations

Research analysis:

You: "What are the key themes in our brand perception survey?"

Claude: I'll analyze that for you.
  → list_surveys (finds "Brand Perception 2025", id: 15)
  → get_response_stats(15) (847 responses)
  → list_questions(15) (finds open-ended Q7)
  → run_theme_extraction(15, question_id: 7)
  → run_sentiment_analysis(15, question_id: 7)

"Here are the 5 main themes from 847 responses..."

Data export:

You: "Export all responses from the employee satisfaction survey as CSV"

Claude:
  → list_surveys (finds "Employee Satisfaction", id: 23)
  → export_responses(23, format: "csv")

"Done — CSV exported with 1,432 responses."

Dashboard creation:

You: "Create a dashboard for the Q1 customer feedback results"

Claude:
  → list_surveys (finds "Q1 Customer Feedback", id: 31)
  → create_dashboard("Q1 Customer Feedback Results")
  → get_response_stats(31)
  → get_summary_statistics(31)

"Dashboard created. Would you like me to add specific charts?"

Best Practices

  • Be specific: "Analyze the NPS question in the Q2 survey" works better than "analyze surveys"
  • Confirm costs: AI analysis operations consume credits. Claude will use them when asked — set boundaries in SKILL.md if needed
  • Workspace context: If you work across multiple workspaces, mention which one
  • Iterative analysis: Start broad (insights, stats) then drill into specific questions