Kibana MCP Server

Name: Kibana MCP Server
Author: ggilligan12

@ggilligan12

代码仓库

关于 Kibana MCP Server

MCP Server for Kibana SecOps

#kibana #mcp #mcp-server

基本信息

分类

其他

许可证

MIT

运行时

python

传输方式

stdio

发布者

ggilligan12

配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "kibana-mcp": {
      "command": "docker",
      "args": [
        "build",
        "-t",
        "kibana-mcp",
        "."
      ]
    }
  }
}

工具

Fetch security alerts

Add tags to alerts

Change alert status (open/acknowledged/closed)

Search detection rules

Get rule exception items

Add exceptions to rules

Create new exception lists

Link exception lists to rules

概览

What is Kibana MCP Server?

A Model Context Protocol (MCP) server that integrates with Kibana Security, enabling AI assistants to manage security alerts, detection rules, and exception lists programmatically.

How to use Kibana MCP Server?

Clone the repository, build the Docker image, set environment variables (KIBANA_URL plus either KIBANA_API_KEY or KIBANA_USERNAME/KIBANA_PASSWORD), then configure your MCP client (e.g., Claude Desktop) to run the container. Alternatively, run locally with uv sync and uv run kibana-mcp.

Key features of Kibana MCP Server

Fetch security alerts via get_alerts
Tag alerts with tag_alert
Change alert status (open/acknowledged/closed)
Search detection rules with find_rules
Retrieve and add rule exception items
Create and associate shared exception lists

Use cases of Kibana MCP Server

Automating alert triage from an AI chat interface
Bulk updating alert statuses during incident response
Adding exception rules to suppress known false positives
Orchestrating rule management across multiple Kibana instances

FAQ from Kibana MCP Server

What authentication methods are supported?

You can authenticate using either a Kibana API key (recommended) or a username/password combination.

How do I configure the server for my MCP client?

Set KIBANA_URL and your credentials as environment variables, then add a Docker-based MCP server entry in your client configuration (e.g., Claude Desktop or Cursor). Direct credentials in the args are also supported.

Can I run the server locally without Docker?

Yes. Install dependencies with uv sync, set the required environment variables, and start the server with uv run kibana-mcp.

Is there a test environment provided?

Yes. The repository includes a testing/quickstart-test-env.sh script that starts a local Kibana/Elasticsearch instance preloaded with test data.

What tools does the server expose?

Eight tools: get_alerts, tag_alert, adjust_alert_status, find_rules, get_rule_exceptions, add_rule_exception_items, create_exception_list, and associate_shared_exception_list.