Project Introduction
I'm excited to announce Kibana MCP Server, a new open-source implementation based on the Model Context Protocol (MCP) that allows any MCP-compatible AI client (such as Claude Desktop) to access and manage Kibana instances through natural language or programmatic requests.
What is MCP?
Model Context Protocol (MCP) is an open protocol developed by Anthropic that enables AI assistants (like Claude) to seamlessly interact with external tools and data sources. Through MCP, AI assistants are no longer limited to their training data but can directly access, query, and operate real-world systems.
Key Features of Kibana MCP Server
-
Natural Language Kibana Access: Directly ask Claude to "create a multi-panel dashboard showing our security posture"
-
Complete Kibana API Coverage: Exposes all Kibana API endpoints as both tools and resources
-
Enterprise Security Features: SSL/TLS support, authentication, custom CA certificates
-
Flexible Configuration: Works with both local and remote Kibana instances
-
Bidirectional Communication: AI can explore data mappings first, then construct precise queries
Typical Use Cases
-
"Create a multi-panel dashboard showing our security posture across all endpoints"
-
"Open case #5021 and summarize all related investigation notes and evidence"
-
"Show me all critical alerts triggered in the last 24 hours and group them by source"
-
"Update the Fleet agent policy to collect additional process metrics on our production servers"
Why We Developed This Project
While there are existing MCP server implementations for Elasticsearch/OpenSearch data access, this implementation specifically focuses on Kibana's interface layer, including dashboards, cases, alerts, and Fleet agent policy management. It complements existing Elasticsearch MCP servers to provide a complete analytics environment where AI can both query raw data and manage its visualization.
Technical Highlights
-
Built with TypeScript using the official MCP SDK
-
Full documentation in both English and Chinese
-
Detailed setup guide and Claude Desktop integration examples
-
Modular design that can run alongside other MCP servers
Try It Now!
GitHub repository: GitHub - TocharianOU/mcp-server-kibana: MCP server for Kibana, Access search and manage Kibana in MCP Client.
Seeking Feedback and Collaboration
As a member of the Elastic community, I'd love to hear your thoughts and suggestions. If you're interested in collaborating to improve the project, please reach out!
-
Which Kibana features do you think would be most useful to access via AI?
-
Have you encountered any issues or have suggestions for improvements?
-
Are you interested in contributing code or documentation?
Thank you for your interest and support!