Kada provides a MCP integration for Claude
Prerequisites
-
You must have access and be licensed for Claude
-
Kada MCP API key (For API) / Secret Value (For Oauth). Oauth is recommended for user MCP access to K.
-
Kada MCP Server URL endpoint (e.g.,
https://[customer].kada.ai/mcp) -
Kada MCP Skills files
-
Discovery Skill
-
Code Extract Skill
-
You must request access to the Kada MCP. Reach out to your Kada account manager for access.
Step 1: Add Kada MCP to Claude connectors
-
Open Claude Desktop. Go to Settings. Go to Developer and Select Edit Config.
-
This opens `claude_desktop_config.json. Edit the config to add the following MCP details
```json { "mcpServers": { "kada-mcp": { "command": "npx", "args": [ "-y", "mcp-remote", "https://[customer].kada.ai/mcp", "--header", "x-api-key:${KADA_MCP_API_KEY}" ], "env": { "KADA_MCP_API_KEY": "<paste key here>" } } } } ```
-
Save the changes to the config file.
-
Restart Claude
-
Go to Settings. Go to Customize. Go to Connectors. The Kada MCP should now show as connected
Step 2: Adding the Kada skills
-
Request the Skills files from your Kada account team
-
Discovery
-
Code extract
-
-
Edit the skills files to reference your Kada domain.
-
Edit the Setup line near the top to reference your Kada domain -
`KADA_DOMAIN = https://[customer].kada.ai`
-
-
Open Claude. Go to Settings. Go to Capabilities. Go to Skills. Upload the Skills files you’ve updated.
Step 3: Testing your Agent
-
Open a new chat
-
Confirm the Kada MCP is connected. Go to Settings. Go to Customize. Go to Connectors. The Kada MCP should now show as connected
-
Ask something — e.g. *"Find tables with customer in the name"*
-
Claude should call `get_search_field_metadata` first, then `search_kada`, and show results with profile links into K
Please advise your Kada account manager with any feedback can be incorporated in the MCP / Skills files in future releases.