UPDATED IN VERSION 6.2
The K MCP brings K into the AI assistants your teams already use. Using the Model Context Protocol (MCP), tools like Claude and Microsoft 365 Copilot can search your data ecosystem, explain assets, extract code and update descriptions in K, all in natural language.
You must request access to the K MCP. Contact your Kada account manager to get started.
Getting started with the K MCP
Setting up the K MCP takes four stages.
1. Request access
Your Kada account manager will provide your MCP server URL, an API key or OAuth client secret, and the skill files you need. OAuth is recommended for individual user access to K.
2. Connect your AI assistant
Add the K MCP as a connector in your AI assistant. Follow the setup guide for Claude or Copilot Studio (for Microsoft 365 Copilot and Teams).
3. Add the skills
Skills tell your assistant how to use K for a specific job. Add the skill for each use case you want to support: Discovery, Update Description or Code Extract.
4. Ask K in natural language
Ask questions or give instructions the way you normally would. K responds with answers and links back to the relevant Data Profiles.
💡 If your assistant gives inaccurate answers, try tuning its instructions. Share feedback with your Kada account manager so it can be built into future skill releases.
Setup guides
|
Guide |
Description |
|---|---|
|
Connect Claude Desktop to K using an API key, then upload the K skills. Requires a Claude licence. |
|
|
Build a K agent in Copilot Studio, connect it to K using OAuth or an API key, and publish it to Microsoft 365 Copilot and Teams. Requires access to Copilot Studio and a Microsoft 365 Copilot licence. |
Use cases
|
Use case |
Skill |
Description |
|---|---|---|
|
Discovery |
Search and explore your data ecosystem in natural language. Find the certified reports in a domain, the best report for an analysis, or the definition of a business term, with links to the relevant profiles. |
|
|
Update Description |
Write and update descriptions for one or many assets and terms, such as every column in a table. Creating or deleting assets and terms, and updating other properties, isn't supported yet. |
|
|
Code Extract |
Extract the code and transformations behind a data asset, such as the SQL that builds a table, and convert it to run on another platform like Databricks. |
Who this is for
Anyone who wants to work with K from their AI assistant. Data consumers use Discovery to find and understand data, Data Owners and Stewards use Update Description to document assets faster, and data engineers use Code Extract to see how assets are built. A K Administrator or IT team member typically sets up the connection.