hypertab API Documentation

The active table where AI agents operate and humans supervise. Columns don't just hold data, they DO things.

AI
For AI Agents
hypertab exposes 49 MCP tools. Connect via the MCP endpoint, read hypertab://tables to see your workspace, and start operating. Every response includes a hints array with the recommended next step.

1. Get your API key

Sign up and open the dashboard MCP page at app.hypertab.ai. It generates a revealed setup key, a full .mcp.json block, and a chat-ready install prompt for your AI agent. Keys start with ht_sk_ and authenticate both REST API and MCP connections.

Your API Key
ht_sk_your_api_key_here

2. Install Hypertab in your agent

Run npx @hypertabai/mcp install once. It writes the MCP config to the right location for your agent and installs the Hypertab skill where supported. No JSON editing, no skill-file hunting, one command and a restart.

Pick your agent on the right. The command prints what it wrote and what to do next.

Prefer env vars on shared machines? Use HYPERTAB_API_KEY=... as a prefix instead of --api-key. Skill file: https://hypertab.ai/skill.md

npx @hypertabai/mcp install
npx @hypertabai/mcp install --client claude-code --scope project --api-key ht_sk_YOUR_KEY

3. Create a table

Tables are the core data structure. Each table has columns (static or smart) and rows. Table and column names must be lowercase with underscores.

MCP Tool Call
{
  "tool": "hypertab_create_table",
  "arguments": {
    "name": "leads",
    "columns": [
      { "name": "company", "type": "text" },
      { "name": "website", "type": "url" },
      { "name": "size", "type": "number" }
    ]
  }
}

4. Insert rows

Insert up to 10,000 rows per call. Column names are fuzzy-matched, if you type companey instead of company, hypertab auto-corrects it and tells you in the response.

MCP Tool Call
{
  "tool": "hypertab_insert_rows",
  "arguments": {
    "table": "leads",
    "rows": [
      { "company": "Acme Corp", "website": "https://acme.com", "size": 500 },
      { "company": "Globex", "website": "https://globex.com", "size": 1200 }
    ]
  }
}

5. Add a smart column

Smart columns process data per row. Use {{column_name}} to reference other columns as input. This HTTP column calls Gemini to classify each lead.

MCP Tool Call
{
  "tool": "hypertab_add_smart_column",
  "arguments": {
    "table": "leads",
    "name": "industry",
    "type": "text",
    "kind": "http",
    "config": {
      "url": "https://generativelanguage.googleapis.com/v1beta/models/gemini-2.5-flash:generateContent?key=YOUR_KEY",
      "method": "POST",
      "headers": { "Content-Type": "application/json" },
      "body": {
        "contents": [{ "parts": [{ "text": "What industry is {{company}} ({{website}})? Reply with one word." }] }]
      },
      "extract": "candidates[0].content.parts[0].text",
      "run_condition": "{{company}} IS NOT EMPTY"
    },
    "rate_limit": "conservative"
  }
}

6. Run the smart column

After adding a smart column, trigger it to process all existing rows. New rows will be processed automatically if auto_run is enabled.

MCP Tool Call
{
  "tool": "hypertab_run_column",
  "arguments": {
    "table": "leads",
    "column": "industry"
  }
}

7. Check progress

Monitor the run status to see how many rows have been processed. The response includes total rows, processed count, error count, and completion percentage.

MCP Tool Call
{
  "tool": "hypertab_get_column_run_status",
  "arguments": {
    "run_id": "run_abc123"
  }
}

8. Query results

Once complete, query the table to see the smart column values populated for each row.

MCP Tool Call
{
  "tool": "hypertab_query_rows",
  "arguments": {
    "table": "leads",
    "columns": ["company", "website", "size", "industry"],
    "limit": 10
  }
}