Kibana

You can connect your MCP-compatible clients to a Kibana instance through this server, enabling natural language or programmatic requests to manage and query Kibana data. It supports multiple transport methods, authentication options, and multi-space Kibana configurations so you can securely operate saved objects and other Kibana resources from your MCP workflow.

How to use

You will run the MCP server in either HTTP mode for remote access or in standard input/output mode for local usage. Your MCP client can then issue requests to Kibana resources such as saved objects, spaces, and API endpoints through this bridge. Use descriptive, natural language queries or structured requests to perform common tasks like listing spaces, searching API paths, or managing dashboards and visualizations.

How to install

Prerequisites you need before installation:

Step-by-step setup using the HTTP transport (remote access) path:

# Install the MCP server globally
npm install -g @tocharianou/mcp-server-kibana

# Or run directly with npx
npx @tocharianou/mcp-server-kibana

# Start with environment variables (example)
MCP_TRANSPORT=http MCP_HTTP_PORT=3000 \
KIBANA_URL=http://localhost:5601 \
KIBANA_API_KEY=your-api-key \
KIBANA_DEFAULT_SPACE=default \
npx @tocharianou/mcp-server-kibana

If you prefer running locally as a stdio server (no remote access):

# Start the server using npx (stdio mode by default)
npx @tocharianou/mcp-server-kibana

# Alternatively, for a pre-built setup with more explicit envs
KIBANA_URL=http://localhost:5601 \
KIBANA_API_KEY=your-api-key \
npx @tocharianou/mcp-server-kibana

Additional configuration and usage notes

Configure authentication and Kibana access using one of the supported methods. The server can operate with an API key, basic authentication, or cookie-based sessions. You can also specify the default Kibana space and manage SSL verification behavior through environment settings.

Remote HTTP mode lets you expose the MCP endpoint at a configurable port. For example, you can access the MCP API at the standard path and perform health checks to ensure the bridge remains available.

Troubleshooting

If you encounter connection or authentication issues, verify that:

• The Kibana URL is reachable from the MCP server host
• Correct authentication credentials are provided (API key, username/password, or cookies)
• SSL validation settings match your environment if you use HTTPS with self-signed certificates
• The MCP HTTP port is accessible if you are using remote access For local development, you can enable verbose debugging to capture request flows and errors for quicker diagnosis.

Debugging

Use the embedded MCP Inspector to troubleshoot interactions between the MCP server and Kibana. Run the inspector to open a browser-based debugging interface that helps you inspect requests, responses, and object state during client operations.

Available tools

get_status

Get Kibana server status via MCP bridge.

execute_kb_api

Execute custom Kibana API requests through the MCP server.

get_available_spaces

List available Kibana spaces accessible through the bridge.

search_kibana_api_paths

Search for Kibana API endpoints via the MCP server.

list_all_kibana_api_paths

List all Kibana API endpoints exposed by the server.

get_kibana_api_detail

Retrieve details about a specific Kibana API path.

vl_search_saved_objects

Search saved objects across Kibana spaces.

vl_get_saved_object

Retrieve a single saved object by ID.

vl_create_saved_object

Create a new saved object in Kibana.

vl_update_saved_object

Update a saved object. Includes partial and full updates.

vl_bulk_update_saved_objects

Bulk update multiple saved objects in one operation.

vl_bulk_delete_saved_objects

Bulk delete multiple saved objects.

analyze_object_dependencies

Analyze dependencies for a given saved object.

analyze_deletion_impact

Check potential impact before deleting a saved object.

check_dashboard_health

Check health status of a dashboard.

scan_all_dashboards_health

Scan all dashboards to assess health issues.