# Pipetable

> Pipetable is an open-source MCP server for files and documents. Local MCP server that lets agents run DuckDB SQL on CSV, Excel, Parquet and JSON files. It runs locally on your machine and is started by your AI client.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** CLI, macOS, Linux, Windows, self-hosted
- **Website:** https://pipetable.com
- **Repository:** https://github.com/melihbirim/pipetable
- **MCP Registry name:** io.github.melihbirim/pipetable
- **Categories:** files and documents, databases, analytics and data
- **Listing updated:** 2026-11-14

## About

Pipetable is built for the case where an assistant needs your business data but you don't want to upload files or open up the whole filesystem. You point it at a folder, it registers the data files it finds, and the agent queries them through DuckDB, so answers come from real query output rather than guesses.

The MCP server exposes four steps: scan a folder, list datasets with column types, inspect one table with sample rows, and execute SQL. Folders are scanned up to three levels deep, hidden files and directories such as node_modules and .git are skipped, and files up to 2GB are supported.

The same binary also works as a standalone CLI with a SQL prompt. Natural-language questions there are optional and use an Anthropic key, an OpenAI-compatible endpoint or a local Ollama model; SQL and the MCP server need no key or internet connection. It is MIT licensed and installs as a small binary or via cargo.

## Features

- **Folder scanning:** Registers CSV, TSV, Parquet, JSON, NDJSON and Excel files found up to three levels deep.
- **Real DuckDB SQL:** Queries run in DuckDB, so results are actual query output.
- **Local-only access:** Files stay on your machine and nothing is uploaded.
- **Schema inspection:** Shows column types and sample rows before the agent writes a query.
- **Standalone CLI:** An interactive prompt with dot commands and one-shot `ask` queries.
- **Chained results:** Each query result is kept as a view called _last for follow-up queries.

## Good for

- Let Claude Code answer questions about a folder of sales CSVs
- Query Excel workbooks without uploading them anywhere
- Explore large Parquet or NDJSON files from an editor assistant
- Run ad hoc SQL over local exports from the terminal

## Install

### Claude Code (Terminal)

```sh
claude mcp add pipetable -- npx -y pipetable
```

### Claude Desktop (claude_desktop_config.json)

```json
{
  "mcpServers": {
    "pipetable": {
      "command": "npx",
      "args": [
        "-y",
        "pipetable"
      ]
    }
  }
}
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "pipetable": {
      "command": "npx",
      "args": [
        "-y",
        "pipetable"
      ]
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "pipetable": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "pipetable"
      ]
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "pipetable": {
      "command": "npx",
      "args": [
        "-y",
        "pipetable"
      ]
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "pipetable": {
      "command": "npx",
      "args": [
        "-y",
        "pipetable"
      ]
    }
  }
}
```

## Tools

- `scan_folder`: Register all data files in a folder
- `list_datasets`: See dataset schemas and column types
- `get_schema`: Inspect a specific table with sample rows
- `execute_sql`: Run DuckDB SQL against the files

## FAQ

### Does it need an API key?

No. SQL and the MCP server work offline without a key; only natural-language questions in the CLI need an Anthropic, OpenAI-compatible or Ollama backend.

### Which file formats can it read?

CSV, TSV, Parquet, JSON, NDJSON and Excel (xlsx, xls, xlsm), up to 2GB per file.

### How do I add it to Claude Code?

Run claude mcp add pipetable pipetable mcp after installing the binary.

### Is my data sent anywhere?

The site and README say files never leave your machine.

### Is Pipetable free?

Yes. Pipetable is open source, so you can use and self-host it for free.

### How do I add Pipetable to Claude Code?

Run this command in your terminal: claude mcp add pipetable -- npx -y pipetable

### What are the best alternatives to Pipetable?

The closest alternatives to Pipetable are Agent-Native Analytics, Google Ads + Meta Ads + SEO/GEO MCP, Tapetide - Indian Stock Market MCP, Midplane and World Monitor.

## Alternatives

- [Agent-Native Analytics](https://ai-powered.com/mcp/agent-native-analytics): Agent-Native Analytics is an open-source MCP server for analytics and data.
- [Google Ads + Meta Ads + SEO/GEO MCP](https://ai-powered.com/mcp/google-ads-meta-ads-seo-geo-mcp): Google Ads + Meta Ads + SEO/GEO MCP is a freemium MCP server for marketing and seo.
- [Tapetide - Indian Stock Market MCP](https://ai-powered.com/mcp/tapetide-indian-stock-market-mcp): Tapetide - Indian Stock Market MCP is a freemium MCP server for finance and payments.
- [Midplane](https://ai-powered.com/mcp/midplane): Midplane is an open-source MCP server for databases.
- [World Monitor](https://ai-powered.com/mcp/world-monitor): World Monitor is a freemium MCP server for analytics and data.
- [CompanyData](https://ai-powered.com/mcp/companydata): CompanyData is a paid MCP server for analytics and data.
- [ASTra](https://ai-powered.com/mcp/astra): ASTra is a freemium MCP server for files and documents.
- [ChartAnt](https://ai-powered.com/mcp/chartant): ChartAnt is a freemium MCP server for design and media.

---
Source: https://ai-powered.com/mcp/pipetable
