Pipetable
About
Pipetable is an MCP server that registers your local data files as DuckDB views and lets AI tools (Claude Code, Cursor, Copilot, RooCode) run real SQL against them. Files never leave your machine. Results are ground truth: not generated.
Details
- Author
- melihbirim
- Downloads
- 338
- Categories
- Database, Other, File Management, Developer Tools
Jump to
- Registers local data files as DuckDB views
- Lets AI tools run real SQL queries
- Files never leave your machine
- Results are ground truth, not generated
- Ships as both MCP server and CLI REPL
- MIT licensed, 5MB binary
Setting up with Highlight
This MCP is not yet compatible with Highlight’s one-click setup. However, you can still use it with Highlight by following these steps:
- Download and install Highlight from highlightai.com/download
- Navigate to the plugins tab and select "Add Custom Plugin"
-
Configure the plugin with the settings below
Plugin Name
PipetableCommand (node, npx, python, etc.)Please refer to the README for specific instructions on how to obtain API keys or other required environment variables.
- Enable "Start Automatically" if you want the plugin to start when Highlight launches
From the repository
You can use Pipetable as an MCP server with supported AI tools or as a standalone CLI REPL. It provides four tools: scan_folder, list_datasets, get_schema, and execute_sql.
Claude Desktop / Cursor
Paste into your MCP client config file to install this server.
{
"mcpServers": {
"pipetable": {
"pipetable": {
"command": "pipetable",
"args": [
"mcp"
]
}
}
}
}McpServers
{
"pipetable": {
"command": "pipetable",
"args": [
"mcp"
]
}
}Point it at a folder of CSV, Parquet, JSON, or TSV files — your AI can now query them with real SQL instead of hallucinating.
Works as an MCP server for Claude Code, Cursor, RooCode, and Copilot. Also ships as a standalone CLI for interactive data exploration. Powered by DuckDB. Files never leave your machine.
# macOS / Linux curl -fsSL https://pipetable.com/install | sh # Windows irm https://pipetable.com/install.ps1 | iex # Rust cargo install pipetable
{ "mcpServers": { "pipetable": { "command": "pipetable", "args": ["mcp"] } } }
{ "servers": { "pipetable": { "type": "stdio", "command": "pipetable", "args": ["mcp"] } } }
- scan_folder— register all data files in a folder
- list_datasets— see schemas and column types
- get_schema— inspect a specific table with sample rows
- execute_sql— run real DuckDB SQL against your files
Results are ground truth from DuckDB, not generated.
SQL and natural language at the>prompt. SQL always works. Natural language requiresOllamarunning locally.
> SELECT region, SUM(revenue) AS total FROM sales GROUP BY 1 ORDER BY 2 DESC 4 row(s) region total ───────────── EU 141000 US 32000 APAC 17000
> show me top 5 customers by revenue Using: customers, sales Thinking..... SELECT c.name, SUM(s.revenue) AS total FROM customers c JOIN sales s ON s.customer_id = c.id GROUP BY c.name ORDER BY total DESC LIMIT 5 ... → piped as _last
Every query saves its result as_last— a live DuckDB view you can query further:
> SELECT FROM sales WHERE region = 'EU' ... → piped as _last > show me top 3 from _last Using: _last Thinking.....
Tab completes dataset names afterFROM,JOIN,.schema,.drop,.use.
pipetable ask "who has the highest revenue?" ~/data/ pipetable ask "SELECT FROM sales LIMIT 5" ~/data/
Set any one of these — pipetable auto-detects:
# Claude (best quality) export ANTHROPIC_API_KEY=sk-ant-... # OpenAI or any compatible API (LM Studio, Groq, Together, etc.) export OPENAI_API_KEY=sk-... export OPENAI_BASE_URL=http://localhost:1234 # optional, for local endpoints # Ollama (local, no key needed) ollama pull qwen2.5-coder:1.5b ollama serve
Priority: Anthropic → OpenAI-compatible → Ollama. SQL and MCP work without any of them.
CSV, Parquet, JSON, NDJSON, TSV, Excel (xlsx, xls, xlsm). Files up to 2GB. Folders scanned up to 3 levels deep. Hidden files and common noise directories (node_modules,target,.git) are skipped automatically.
Query a folder of CSV / Parquet / JSON files with SQL via DuckDB — read-only and sandboxed; scattered files become one queryable source.
Official MCP server for dbt (data build tool) providing integration with dbt Core/Cloud CLI, project metadata discovery, model information, and semantic layer querying capabilities.
Query and analyze data with MotherDuck and local DuckDB
A collection of tools for managing the platform, addressing data quality and reading and writing to Teradata Database.
A read-only MCP server for Avro data sources, powered by the CData JDBC Driver.
Run SQL queries on data in Amazon S3 using AWS Athena.
Interact with Bauplan data tables and run queries.
Production-grade MCP server for Databricks: SQL Warehouses, Jobs API, multi-workspace support.
Visual no-code generator that turns any database into multiple scoped MCP servers — one per access group, with PII masking and fail-closed query scoping built in.
A read-only MCP server by CData that enables LLMs to query live data from EnterpriseDB databases.
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.
