dlt
Official<h1 align="center">
<strong>data load tool (dlt) — MCP Server</strong>
</h1>
<p align="center">
🚀 Follow <a href="https://dlthub.com/docs/dlt-ecosystem/llm-tooling/llm-native-workflow">this guide</a> to create a dlt pipeline in 10mins with AI
</p>
## How is it useful?
Large language models (LLMs) know a lot about the world, but nothing about your specific code and data.
The [Model Context Protocol](https://modelcontextprotocol.io/) (MCP) server allows the LLM to retrieve **up-to-date** and **correct** information about your [dlt](https://github.com/dlt-hub/dlt) pipelines, datasets, schema, etc. This significantly improves the development experience in AI-enabled IDEs (Copilot, Cursor, Continue, Claude Code, etc.)
## Installation
The package manager [uv](https://docs.astral.sh/uv/getting-started/installation/) is required to launch the MCP server.
Add this section to your MCP configuration file inside your IDE. Add your destination(s) in the extras `dlt-mcp[...]`
```json
{
"name": "dlt",
"command": "uv",
"args": [
"run",
"--with",
"dlt-mcp[duckdb]",
"dlt-mcp",
],
}
```
>[!NOTE]
>The configuration file format varies slightly across IDEs
## Features
### Tools
The dlt MCP server provides [tools](https://modelcontextprotocol.io/specification/2025-11-25/server/tools) that allows the LLM to take actions:
- **list_pipelines**: Lists all available dlt pipelines. Each pipeline consists of several tables.
- **list_tables**: Retrieves a list of all tables in the specified pipeline.
- **get_table_schemas**: Returns the schema of the specified tables.
- **execute_sql_query**: Executes a SELECT SQL statement for simple data analysis.
- **get_load_table**: Retrieves metadata about data loaded with dlt.
- **get_pipeline_local_state**: Fetches the state information of the pipeline, including incremental dates, resource state, and source state.
- **get_table_schema_diff**: Compares the current schema of a table with another version and provides a diff.
- **search_docs**: Searches over the `dlt` documentation using different modes (hybrid, full_text, or vector) to verify features and identify recommended patterns.
- **search_code**: Searches the source code for the specified query and optional file path, providing insights into internal code structures and patterns.
TDQS
Scored across 8 tools
Each tool has a distinct purpose: display_schema for diagram, execute_sql_query for running SQL, get_load_table for load metadata, get_pipeline_local_state for state, get_table_schema for schema, get_table_schema_diff for version diff, list_pipelines for pipeline listing, list_tables for table listing. No overlap.
All tools follow a consistent verb_noun pattern with snake_case (e.g., display_schema, get_table_schema, list_pipelines). No mixing of conventions.
With 8 tools covering pipeline inspection, schema viewing, state retrieval, and query execution, the count is well-scoped for a data pipeline management server.
The tool set covers inspection and querying comprehensively (list, schema, state, load info, SQL execution), but lacks lifecycle operations like creating or deleting pipelines, which may be intentional for a read-only server.