Kedro MCP Server
OfficialEnables conversion of Jupyter notebooks into production-ready Kedro projects, including analysis, planning, and automated scaffolding.
Provides guidance and tools for working with Kedro projects, including general usage, best practices, project migration between versions, and converting Jupyter notebooks into Kedro projects.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Kedro MCP Server migrate my Kedro project from 0.19 to 1.0"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Kedro MCP Server
An MCP (Model Context Protocol) server that helps AI assistants (such as VS Code Copilot or Cursor) work consistently with Kedro projects.
The server provides concise, versioned guidance for:
General Kedro usage and best practices
Converting Jupyter notebooks into production-ready Kedro projects
Migrating projects between Kedro versions
With Kedro-MCP, your AI assistant understands Kedro workflows, pipelines, and conventions — so you can focus on building, not fixing AI mistakes.
Quick Install
To enable Kedro MCP tools in your editor, simply click one of the links below.
Your editor will open automatically, and you’ll just need to confirm installation.
Once installed, your AI assistant automatically gains access to Kedro-specific MCP tools.
Helpful references
Universal MCP configuration (JSON)
You can reuse this configuration in any MCP-compatible client (e.g. Copilot, Cursor, Claude, Windsurf):
{
"command": "uvx",
"args": ["kedro-mcp@latest"],
"env": {
"FASTMCP_LOG_LEVEL": "ERROR"
},
"disabled": false,
"autoApprove": []
}Related MCP server: CastPlan MCP
Usage
After installation, open Copilot Chat (in Agent Mode) or the Chat panel in Cursor.
Type / to see available Kedro MCP prompts.
Convert a Jupyter Notebook into a Kedro project
/mcp.Kedro.convert_notebookWhen you run this command, the assistant explicitly calls the Kedro MCP server and follows the guidance provided.
Typical flow:
The assistant analyses your Jupyter notebook (you can paste its content or mention its filename).
It creates a conversion plan (Statement of Work) saved as a
.mdfile in your workspace.You review and approve the plan.
The assistant:
Ensures a Python virtual environment is active.
Installs the latest Kedro if missing.
Scaffolds a new project with
kedro new.Creates pipelines with
kedro pipeline create.Populates
parameters.ymlandcatalog.ymlbased on your notebook.
You can edit the plan, switch environment tools (uv, venv, conda), or ask the assistant to resolve setup errors interactively.
Migrate a Kedro project
/mcp.Kedro.project_migrationThis prompt walks you through migrating an existing Kedro project to a newer version.
Steps:
The assistant analyses your project and proposes a migration plan (e.g. from 0.19 → 1.0).
You review and approve the plan.
The assistant ensures a virtual environment is active, installs the correct Kedro version, and applies migration steps.
Use this to get up-to-date migration tips and avoid deprecated patterns.
General Kedro guidance
/mcp.Kedro.general_usageUse this prompt for open-ended Kedro questions.
The Kedro MCP server returns structured, up-to-date Kedro guidance that your assistant uses to generate realistic code and pipelines.
Example:
“Generate a Kedro project for a time-series forecasting pipeline using Pandas and scikit-learn.”
Manual Install (from source)
For development or debugging:
git clone https://github.com/kedro-org/kedro-mcp.git
cd kedro-mcp
uv pip install -e . --group devExample MCP config (local path):
{
"mcpServers": {
"kedro": {
"command": "uv",
"args": ["tool", "run", "--from", ".", "kedro-mcp"],
"env": { "FASTMCP_LOG_LEVEL": "ERROR" }
}
}
}Development
# Install dev dependencies
uv pip install -e . --group dev
# Lint & type-check
ruff check .
mypy src/Troubleshooting
Server not starting: Ensure Python 3.10+ and
uvare installed. Confirm the MCP config points touvx kedro-mcp@latestor to thekedro-mcpconsole script.Tools not appearing: Restart your assistant and verify that the MCP config key matches
"kedro".Version drift: Pin a version instead of
@latestfor reproducibility.
License
This project is licensed under the Apache Software License 2.0.
See LICENSE.txt for details.
Support
Report issues: https://github.com/kedro-org/kedro-mcp/issues
Learn more about MCP: https://modelcontextprotocol.io/
Available Tools
3 toolskedro_general_instructionsA
Return general Kedro usage guidance.
Inputs: no named parameters. Returns: result
notebook_to_kedroB
Return Notebook→Kedro conversion instructions.
Inputs: no named parameters. Returns: result
project_migrationC
Return project migration instructions.
Inputs: no named parameters. Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 3 tools
Each tool clearly targets a distinct area: general usage, notebook conversion, and project migration. There is no overlap in their described purposes, making selection unambiguous.
Names are readable but follow different conventions: 'kedro_general_instructions' uses a prefix, 'notebook_to_kedro' uses an arrow-style descriptor, and 'project_migration' is a plain noun phrase. The lack of a uniform pattern is noticeable.
The server is narrowly scoped to providing instructions, and three tools cover its intended sub-areas. While not excessive, the set is minimal and could benefit from a few more topics to feel more complete.
The domain is Kedro, yet the tool surface covers only general guidance, notebook conversion, and project migration. Missing are tools for pipeline development, data catalog management, or execution, which are core Kedro workflows. This leaves significant gaps for agents needing practical help.
Maintenance
Related MCP Connectors
Versioned documentation registry and semantic search for AI tools and coding assistants.
Project memory, documentation and controlled secret access for AI coding agents.
Gives AI agents direct access to your Sanity projects and guidance for working with them
Serves your design system and coding standards to coding agents, so they stop guessing.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceProvides code manipulation, execution, and version control capabilities. It allows AI assistants to read, write, and execute code while maintaining a history of changes.9MIT
- AlicenseBqualityCmaintenanceProvides AI assistants with persistent memory of your project architecture, development history, and technical decisions, allowing them to give context-aware coding help without needing repeated explanations.1661 npm2MIT
- AlicenseNot gradedqualityDmaintenanceProvides AI agents with queryable, version-controlled project rules and coding standards. Enables validation, rule-based guidance, and task summaries to keep AI work aligned with your project's conventions without repeating context.2MIT
- AlicenseBqualityDmaintenanceProvides AI assistants with accurate Flyte V2 knowledge, patterns, and plugin information to help developers write correct Flyte code. It enables tasks like learning the V2 API, finding examples, selecting plugins, and migrating V1 code to V2.17Apache 2.0