# Karini MCP Server
## **karini-mcp-server**
Karini's native MCP Server, which exposes agents and APIs through standard MCP clients. With this launch, you can authenticate using MCP clients embedded within your favorite tools and access authorized endpoints:
**1. List and query copilots** - Copilots can expose multi-agentic workflows
**2. Manage webhook agents** - List long-running agents, invoke them, and query execution status
**3. Access execution traces** - Retrieve traces for specific executions to troubleshoot runtime issues
**4. Search knowledge bases** - Perform searches across Karini's native Knowledge Bases
These capabilities unlock powerful use cases while respecting API access controls.
These MCP server enables seamless integration with multiple AI development platforms. This allows your AI agents and copilots to access Karini AI's capabilities directly from your preferred development environment.
#### QuickSuite with Karini MCP Server
This setup reference covers setting up the Karini MCP (Model Context Protocol) Server integration with Amazon QuickSuite, enabling AI-powered agents to interact with your Karini AI platform.
#### **Prerequisites**
* Access to Amazon QuickSuite.
* Active Karini AI account with API access.
* Admin permissions in your QuickSuite organization.
#### **Setup Steps**
1. **Navigate to Integrations**
1. Open Amazon QuickSuite → **Integrations** → **Actions** tab.
2. Click **+** on the **Model Context Protocol** card.
2. Connect - Basic Configuration
1. **Name:** Provide a descriptive name to help chat agents find and use this integration.
2. **Description:** Using this MCP server you can access agents build in karini ai agentic platform.
3. **MCP Server Endpoint:** `https://app.karini.ai/api/mcp/sse`
**Note: If you are using your own private Karini AI domain then MCP Server Endpoint should be https\://\/api/mcp/sse**
4. **Auto-publishing:** Enable.
5. Click **Next**.
3. Authenticate
1. **Method:** User authentication.
2. **Organization:** Select your organization when prompted.
3. Click **Create and continue**
4. Available Karini AI MCP actions:
* `invoke_copilot` - Invoke a copilot with a question
* `list_copilots` - List all available copilots
* `search_tool` - Search within a dataset
* `list_datasets` - List all available datasets
* `invoke_webhook` - Invoke a webhook
* `list_webhooks` - List configured webhooks
* `get_traces` - Get traces for a request
* `get_webhook_status` - Get webhook status
Click **Next**.
5. Share (Optional)
1. Add users/groups for shared access, or click **Done** to complete setup.
6. Once set up, chat agents in QuickSuite can automatically invoke Karini AI capabilities:
* Query copilots through natural language
* Search datasets
* Invoke webhooks for automated workflows
* Retrieve trace information for debugging
Refer the video for setting up the MCP server integration with Amazon QuickSuite.
{% embed url="" %}
#### Cursor with Karini MCP Server
This setup reference covers setting setting up the Karini MCP Server integration with Cursor IDE, enabling AI-powered coding assistance with direct access to your Karini AI copilots, datasets, and webhooks.
**Prerequisites**
* Cursor IDE installed
* Active Karini AI account with API access
* Access to `.cursor` configuration directory
**Setup Steps**
1. **Open Cursor Settings**
1. Open Cursor IDE
2. Press `Cmd+,` (Mac) or `Ctrl+,` (Windows/Linux) to open Settings.
3. Navigate to **Tools & MCP** in the left sidebar
2. **Locate MCP Configuration File**
You can access this file by clicking on the `mcp.json` file in the breadcrumb at the top of the editor.
3. **Configure MCP Server**
Add the following configuration to your `mcp.json` file:
```json
{
"mcpServers": {
"Karini-MCP": {
"url": "https://app.karini.ai/api/mcp/sse"
}
}
}
```
4. Connect to MCP Server
1. Go back to **Cursor Settings** → **Tools & MCP**
2. Under **Installed MCP Servers**, locate **Karini-MCP**
3. Click **Connect** button
4. Status will show "Needs authentication"
5. Authenticate
1. Click **Connect** again to initiate OAuth flow.
2. Cursor will prompt: "Do you want Cursor to open the external website?"
1. URL format: `.../app.karini.ai/api/oauth/authorize/response_type=code&client_id=...`
3. Click **Open** to proceed
4. Browser opens showing "Verifying your account...".
6. After login
1. Select your organization when prompted
2. Shows: "Logged in as: \[]"
3. "Completing authentication..." message displays
4. Return to Cursor IDE
7. Verify Connection
1. Back in **Cursor Settings** → **Tools & MCP**
2. **Karini-MCP** should show:Status: "8 tools enabled".
8. Available Karini AI MCP actions:
* `invoke_copilot` - Invoke a copilot with a question
* `list_copilots` - List all available copilots
* `search_tool` - Search within a dataset
* `list_datasets` - List all available datasets
* `invoke_webhook` - Invoke a webhook
* `list_webhooks` - List configured webhooks
* `get_traces` - Get traces for a request
* `get_webhook_status` - Get webhook status
9. **Using the Integration-In Cursor Chat**
1. Open a new chat in Cursor (right panel)
2. Type requests like
1. "List down available tools for Karini-MCP".
2. "Search the customer dataset for recent complaints".
3. "What copilots are available?"
4. "Invoke the sales assistant copilot"
Refer the video for setting up the MCP server integration with Cursor IDE.
{% embed url="" %}
#### Claude with Karini MCP Server
This setup reference covers setting up the Karini MCP Server integration with Claude , enabling AI assistance with direct access to your Karini AI copilots, datasets, and webhooks.
**Prerequisites**
* Access to Claude in a browser (`claude.ai`)
* Active Karini AI account with API/MCP access
* Access to Connections / MCP settings in the Claude UI
**Setup Steps**
1. **Open Claude Settings**
1. Open Claude (`claude.ai`).
2. Open **Settings** from the left navigation.
3. Navigate to **Connectors.**
2. Add Custom Connector
1. On the **Connectors** page, click **Add custom connector.**
2. A modal opens: **Add custom connector.**
3. Configure MCP Server
1. Add the following server configuration:
1. **Name:** `Karini-MCP`
2. **Remote MCP server URL**
```
https://app.karini.ai/api/mcp/sse
```
2. Click **Add**..
4. Configure Tool Permissions
1. Click "Configure" next to the new karini-mcp connector
2. Review tool permissions:
1. **Write/delete tools** (3 tools)
2. **Other tools** (5 tools)
3. Available tools include:`get_traces` -
1. Get traces for request ID
2. `get_webhook_status` - Get webhook status
3. `invoke_copilot` - Invoke copilot with question
4. `invoke_webhook` - Invoke webhook with input
5. `list_copilots` - List available copilots
6. `list_datasets` - List available datasets
5. Test Connection
1. Return to chat interface
2. Ask Claude to list karini-mcp tools
3. Verify tools are accessible and functional
Refer the video for setting up the MCP server integration with Claude.
{% embed url="" %}
## karini-sap-odata-mcp and karini-sap-mcp
* Acts as a standardized integration layer for accessing **SAP HANA** databases and **SAP OData** endpoints.
* Enables tools and prompt workflows to **discover available tables and entities**.
* Supports schema and **metadata inspection**, including details such as **columns, data types, and relationships**.
* Allows execution of **SQL queries against SAP HANA**.
* Provides a **consistent MCP interface** for querying and exploring SAP data sources, eliminating the need for direct client-side connectivity or custom integration logic.
**Tools for ODATA:**
| **Tool Name** | **Description** |
| ---------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| list\_odata\_tables | Lists all available OData tables/entities from the connected service and returns metadata such as catalog, schema, table/entity name, type, and updateability. |
| get\_odata\_table\_schema | Retrieves detailed schema information for a specified OData table, including column names, data types, sizes, nullable status, and key information. |
| execute\_odata\_query | Executes SQL queries against the SAP OData service. Supports SELECT, INSERT, UPDATE, and DELETE operations. Queries can be executed with or without schema prefixes; double quotes should be used for table names containing special characters. |
| count\_odata\_table\_records | Returns the number of records in an OData table, with optional filtering. Filtering conditions should be provided without the WHERE keyword. |
**Tools for SAPHANA**
| **Tool Name** | **Description** |
| ------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| list\_saphana\_tables | Lists all available SAP HANA tables with optional schema filtering and returns metadata such as catalog, schema, table name, and type. |
| get\_saphana\_table\_schema | Retrieves detailed schema information for a specified SAP HANA table, including column names, data types, sizes, and nullable status. |
| execute\_saphana\_query | Executes SQL queries against the SAP HANA database. Supports SELECT, INSERT, UPDATE, and DELETE operations and requires fully qualified table names with proper quoting. |
| count\_saphana\_table\_records | Returns the number of records in a SAP HANA table with optional filtering. Filtering conditions should be provided without the WHERE keyword. |
Limitation: The current Karini AI MCP server only supports Google Auth and will soon start supporting other OAuth providers such as Azure, Okta and native Auth
\
\
---
# Agent Instructions: Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter:
```
GET https://karini-ai.gitbook.io/karini-ai-documentation/mcp/karini-mcp-server.md?ask=
```
The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.