MCP Server Documentation — SPSS InsightGenius

Connecting Claude Desktop

Once connected, you can chat with your SPSS data inside Claude Desktop. All 14 analysis tools become available — frequency tables, crosstabs with significance, NPS, full tabulations exported to Excel — with no code.

Get an API key. Generate one at spss.insightgenius.io/admin/keys (Free tier works — 100 respondents/file cap).
Open claude_desktop_config.json:
- macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
- Windows: %APPDATA%\Claude\claude_desktop_config.json
- Linux: ~/.config/Claude/claude_desktop_config.json
If the file doesn't exist, create it. If it already has mcpServers, add the entry below alongside the others.

Paste this snippet (replace YOUR_INSIGHTGENIUS_API_KEY with your key from step 1):

{
  "mcpServers": {
    "insightgenius": {
      "url": "https://spss.insightgenius.io/mcp/sse",
      "headers": {
        "Authorization": "Bearer YOUR_INSIGHTGENIUS_API_KEY"
      }
    }
  }
}

Save and fully quit Claude Desktop (Cmd/Ctrl+Q — not just closing the window). Reopen it.
Verify the connection. Click the hammer icon (🔨) below the chat input — you should see insightgenius listed with 14 tools available. Try:
"Upload my survey.sav, then show me the frequency of Q1 broken down by gender with significance letters."

Troubleshooting: If Claude Desktop reports "Failed to connect," double-check the URL ends in /mcp/sse (not /mcp/http), your API key is valid (test it at /docs), and you fully quit + reopened the app.

Other Clients

Cursor IDE

Same JSON snippet as above, pasted into Cursor's MCP settings (Settings → MCP Servers → Add new MCP server).

Claude.ai Web (OAuth Connector)

Connect via Settings → Connectors → "Add custom connector" → URL: https://spss.insightgenius.io/mcp/sse. Authenticates via OAuth — no API key needed.

n8n / Make / Zapier (Automation)

SSE Endpoint: https://spss.insightgenius.io/mcp/sse
Header: Authorization: Bearer YOUR_INSIGHTGENIUS_API_KEY

How It Works

1

Upload

spss_upload_file stores your file in a 30-minute session. Returns a file_id.

2

Analyze

Use file_id with any analysis tool. No re-uploading. Get structured results + insights.

3

Present

Every response includes content_blocks ready for Gamma, PowerPoint, Canva, or Google Slides.

Tools (13)

System

spss_get_server_info

Engine status, available tools, plan limits, supported formats. No auth required.

spss_upload_file writes

Upload .sav/.csv/.xlsx file. Returns file_id for reuse (30-min TTL, refreshes on use).

Exploration

spss_get_metadata

Variable names, labels, types, AI-detected: suggested banners, groups (MRS/Grid), preset nets.

spss_describe_variable

Deep profile of one variable: distribution, labels, missing values, mean/std/median.

Analysis

spss_analyze_frequencies

Frequency tables for 1-50 variables in one call (batch). Counts, %, mean, std, median.

spss_analyze_crosstab

Cross-tabulation with significance letters (A/B/C), chi-square p-value, optional means + T-test.

spss_analyze_correlation requires MRX

Correlation matrix (Pearson/Spearman/Kendall) with p-values and significance flags.

spss_analyze_anova requires MRX

One-way ANOVA with Tukey HSD post-hoc pairwise comparisons.

spss_analyze_gap requires MRX

Importance-Performance gap analysis with priority classification and quadrant assignment.

spss_summarize_satisfaction requires MRX

Compact summary: Top 2 Box %, Bottom 2 Box %, Mean for multiple scale variables in one call.

Output

spss_auto_analyze zero-config

Upload file, get complete Excel. AI auto-detects banners, groups, nets. Returns download URL.

spss_create_tabulation

Full professional Excel: sig letters, nets, means, MRS, grids, custom groups. Returns download URL + tables_summary.

spss_export_data

Convert data to xlsx, csv, parquet, or dta format. Returns download URL.

Response Format

Every analysis tool returns a standard envelope with structured data, a deterministic insight summary, and presentation-ready content blocks:

{
  "tool": "spss_analyze_frequencies",
  "file_id": "abc123",
  "variables_analyzed": ["Q1_satisfaction"],
  "sample_size": 796,
  "results": [ ... ],
  "insight_summary": "Q1: most common is 'Satisfied' (38.2%, n=796)",
  "content_blocks": {
    "title": "Customer Satisfaction Distribution",
    "key_finding": "'Satisfied' is the most common response (38.2%)",
    "data_table_markdown": "| Value | % | Count |..."
  }
}

content_blocks is tool-agnostic — feed it directly to Gamma, PowerPoint, Canva, or Google Slides.

File Sessions

Upload once, analyze many times. No need to re-send 50MB of data on every call.

1. spss_upload_file(file_base64="...", filename="survey.sav")
   → { file_id: "abc123", ttl_seconds: 1800 }

2. spss_get_metadata(file_id="abc123")
   → { variables: [...], suggested_banners: [...] }

3. spss_analyze_frequencies(file_id="abc123", variables=["Q1", "Q2"])
   → { results: [...], insight_summary: "..." }

Sessions last 30 minutes (sliding window — resets on each use).

Backwards compatible: you can still pass file_base64 directly to any tool.

Supported Formats

Format	Extension	Metadata
SPSS	.sav, .por, .zsav	Full (variable labels, value labels, types)
CSV	.csv, .tsv	Inferred from column names + dtypes
Excel	.xlsx, .xls	Inferred from column names + dtypes

Authentication

All tools require an API key (sk_live_... or sk_test_...).

Streamable HTTP: Pass as Authorization: Bearer YOUR_KEY header
SSE: Pass as api_key parameter in each tool call

Data Privacy

Stateless: Files processed in-memory, never stored on disk
Sessions: File data in Redis with 30-min auto-delete TTL
Downloads: 5-min TTL, then permanently deleted
AI: Only metadata/labels sent to Haiku — never individual responses
No retention: No database, no backups, no data sharing

Full policy: spss.insightgenius.io/privacy