About Dlt MCP
Dlt MCP is an MCP server published by dlt-hub in the Developer Tools category: mCP server for the Python library dlt.
The repository has 10 stars and 3 forks, with the last commit 8 months ago. Six months or more without a commit doesn't mean the server is broken, but check the open issues before depending on it in production.
Install
git clone https://github.com/dlt-hub/dlt-mcpThis server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.
Tools (9)
execute_sql_queryExecutes a SELECT SQL statement for simple data analysis.
Parameters not published.
get_load_tableRetrieves metadata about data loaded with dlt.
Parameters not published.
get_pipeline_local_stateFetches the state information of the pipeline, including incremental dates, resource state, and source state.
Parameters not published.
get_table_schema_diffCompares the current schema of a table with another version and provides a diff.
Parameters not published.
get_table_schemasReturns the schema of the specified tables.
Parameters not published.
list_pipelinesLists all available dlt pipelines. Each pipeline consists of several tables.
Parameters not published.
list_tablesRetrieves a list of all tables in the specified pipeline.
Parameters not published.
search_codeSearches the source code for the specified query and optional file path, providing insights into internal code structures and patterns.
Parameters not published.
search_docsSearches over the `dlt` documentation using different modes (hybrid, full_text, or vector) to verify features and identify recommended patterns.
Parameters not published.
Ask AI
Ask AI about Dlt MCP
Powered by Claude · Grounded in docs
Security checks
Static checks derived from the repository 4 days ago. Heuristics on the source, not an audit.
- ✓README presentThe repository ships documentation.
- ✓License declaredUsable under a stated license.
- ✓Tests presentA test suite exists in the repository.
- ✓Dependencies pinnedA lockfile pins dependency versions.
- ✓No dynamic code executionSource scan found no eval, dynamic Function, or shell execution.
- !Scoped permissionsDoesn't declare a permission scope. Assume it can do anything its process can.
Releases
README
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 (MCP) server allows the LLM to retrieve up-to-date and correct information about your 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 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[...]
{
"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 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
dltdocumentation 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.
README fetched from the repository 4 days ago. The original is authoritative.