Kibana MCP Server
TocharianOU/mcp-server-kibana · 84 stars · TypeScript · Apache-2.0
MCP server MCP server for Kibana, Access search and manage Kibana in MCP Client.
Install
The repo has no one-line install. Follow its README.
Published as
- npm
@tocharianou/mcp-server-kibanastdio · Streamable HTTP
From the server's entry in the official MCP Registry.
Configuration
| Name | Set as | What it is |
|---|---|---|
KIBANA_URLRequired | Env var | Your Kibana instance URL (e.g., http://localhost:5601) |
KIBANA_API_KEYOptional · Secret | Env var | Kibana API key for authentication |
KIBANA_USERNAMEOptional | Env var | Kibana username for basic authentication |
KIBANA_PASSWORDOptional · Secret | Env var | Kibana password for basic authentication |
MCP_TRANSPORTRequired | Env var | Transport mode selection |
MCP_HTTP_PORTOptional | Env var | HTTP server port |
MCP_BEARER_TOKENOptional · Secret | Env var | Bearer token required on /mcp requests (recommended for any non-localhost bind) |
Files
Kibana MCP Server
A Kibana MCP server implementation that allows any MCP-compatible client (such as Claude Desktop) to access your Kibana instance via natural language or programmatic requests.
This project is based on the official Elastic Kibana API documentation and uses the OpenAPI YAML specification from Elastic Stack 8.x. For details, see the Kibana API documentation.
This project is community-maintained and is not an official product of Elastic or MCP.
💡 Companion Project: For complete Elastic Stack integration, pair this with Elasticsearch MCP Server for direct Elasticsearch data operations.
🚀 Installation
# Global installation (recommended) npm install -g @tocharianou/mcp-server-kibana # Or use directly with npx npx @tocharianou/mcp-server-kibana
From Source
git clone https://github.com/TocharianOU/mcp-server-kibana.git cd mcp-server-kibana npm install && npm run build
🎯 Quick Start
Claude Desktop Integration (Recommended)
Add to your Claude Desktop configuration file:
- MacOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"kibana": {
"command": "npx",
"args": ["@tocharianou/mcp-server-kibana"],
"env": {
"KIBANA_URL": "http://your-kibana-server:5601",
"KIBANA_API_KEY": "your-api-key",
"KIBANA_DEFAULT_SPACE": "default"
}
}
}
}
Direct CLI Usage
# Using API Key (recommended) KIBANA_URL=http://localhost:5601 \ KIBANA_API_KEY=your-api-key \ npx @tocharianou/mcp-server-kibana # Using Basic Auth KIBANA_URL=http://localhost:5601 \ KIBANA_USERNAME=your-username \ KIBANA_PASSWORD=your-password \ npx @tocharianou/mcp-server-kibana # Using Cookie Auth KIBANA_URL=http://localhost:5601 \ KIBANA_COOKIES="sid=xxx; security-session=yyy" \ npx @tocharianou/mcp-server-kibana
HTTP Mode (Remote Access)
MCP_TRANSPORT=http \ MCP_HTTP_PORT=3000 \ KIBANA_URL=http://localhost:5601 \ KIBANA_API_KEY=your-api-key \ npx @tocharianou/mcp-server-kibana
Access at: http://localhost:3000/mcp
Health check: http://localhost:3000/health
Securing HTTP mode. Every session on /mcp acts with the Kibana credentials the server was started with, so treat the endpoint as sensitive:
- By default the server listens on
localhostonly and rejects requests whoseHostheader is not a loopback name (DNS rebinding protection). - To accept remote clients, set
MCP_BEARER_TOKEN. Clients must then sendAuthorization: Bearer <token>on every/mcprequest./healthstays open. - Binding to a non-loopback address (
MCP_HTTP_HOST=0.0.0.0) withoutMCP_BEARER_TOKENlogs a warning at startup: anyone who can reach the port can use your Kibana credentials. - Behind a reverse proxy that forwards its public
Hostheader to a localhost-bound server, list that hostname inMCP_ALLOWED_HOSTS.
MCP_TRANSPORT=http \ MCP_HTTP_HOST=0.0.0.0 \ MCP_BEARER_TOKEN="$(openssl rand -hex 32)" \ KIBANA_URL=http://localhost:5601 \ KIBANA_API_KEY=your-api-key \ npx @tocharianou/mcp-server-kibana
✨ Features
Core Capabilities
- Dual transport modes: Stdio (local) and HTTP (remote access)
- Multiple authentication methods: API Key, Basic Auth, Cookie-based
- Multi-space support: Enterprise-ready Kibana space management
- SSL/TLS support: Custom CA certificate configuration
- Session management: Automatic UUID generation for HTTP mode
- Dynamic API discovery: Based on official Kibana OpenAPI specification
Saved Objects Management
- Complete CRUD operations for all Kibana saved object types
- Intelligent search with pagination support
- Bulk operations for efficient mass updates
- Version control with optimistic concurrency
- Reference management for object relationships
🔧 Configuration
Required Variables
| Variable | Description | Example |
|---|---|---|
KIBANA_URL |
Kibana server address | http://localhost:5601 |
Authentication (choose one method)
| Variable | Description | Priority |
|---|---|---|
KIBANA_API_KEY |
API Key (base64 encoded) | 1st |
KIBANA_USERNAME + KIBANA_PASSWORD |
Basic authentication | 2nd |
KIBANA_COOKIES |
Session cookies | 3rd |
Optional Variables
| Variable | Description | Default |
|---|---|---|
KIBANA_DEFAULT_SPACE |
Default Kibana space | default |
KIBANA_CA_CERT |
CA certificate path | - |
KIBANA_TIMEOUT |
Request timeout (ms) | 60000 |
MCP_TRANSPORT |
Transport mode | stdio |
MCP_HTTP_PORT |
HTTP server port | 3000 |
MCP_HTTP_HOST |
HTTP server host | localhost |
MCP_BEARER_TOKEN |
HTTP mode: require Authorization: Bearer <token> on /mcp |
- |
MCP_ALLOWED_HOSTS |
HTTP mode: extra hostnames accepted in the Host header (comma-separated) |
- |
NODE_TLS_REJECT_UNAUTHORIZED |
Disable SSL validation | 1 |
🛠️ Available Tools
Base Tools
get_status- Get Kibana server statusexecute_kb_api- Execute custom Kibana API requestssearch_elasticsearch- Search Elasticsearch documents and logs through Kibana's search API, within the user's Elasticsearch index privilegesget_available_spaces- List available Kibana spacessearch_kibana_api_paths- Search API endpointslist_all_kibana_api_paths- List all API endpointsget_kibana_api_detail- Get API endpoint details
Saved Objects Tools
vl_search_saved_objects- Search saved objects (universal)vl_get_saved_object- Get single saved objectvl_create_saved_object- Create new saved objectvl_update_saved_object- Update single saved objectvl_bulk_update_saved_objects- Bulk update operationsvl_bulk_delete_saved_objects- Bulk delete operations
Supported Object Types: dashboard, visualization, index-pattern, search, config, lens, map, tag, canvas-workpad, canvas-element
Analysis Tools (v0.6.0+)
analyze_object_dependencies- Analyze saved object dependenciesanalyze_deletion_impact- Check impact before deletioncheck_dashboard_health- Dashboard health checkscan_all_dashboards_health- Batch health scanning
Elasticsearch Data Search
Use search_elasticsearch when you need to search documents, application logs, traces, or aggregations in Elasticsearch. It sends Query DSL through Kibana's /internal/search/es route instead of the Dev Tools Console proxy.
The tool is read-only and does not widen data access: Elasticsearch enforces the configured user's index privileges on every search, exactly as it does for Discover and dashboards. Indices the user cannot read return no results.
It exists because the Dev Tools Console proxy (/api/console/proxy) is gated by a separate Kibana privilege, since the Console can call any Elasticsearch API including write and admin endpoints. A user who can read the target indices but has no Dev Tools privilege gets 403 from the Console proxy; this tool gives that user search, and nothing more. If your Kibana administrators rely on withholding Dev Tools to block ad-hoc queries from an account, be aware that this tool allows Query DSL searches for it.
Example tool arguments for a bounded log search:
{
"index": "application-logs-*",
"body": {
"size": 20,
"_source": [
"@timestamp",
"message",
"trace.id",
"service.name"
],
"sort": [
{ "@timestamp": "desc" }
],
"query": {
"bool": {
"filter": [
{
"range": {
"@timestamp": {
"gte": "2025-01-15T10:00:00Z",
"lt": "2025-01-15T11:00:00Z"
}
}
},
{
"term": {
"service.name.keyword": "example-service"
}
}
]
}
}
}
}
For high-volume logs, start with a small time window and a small size. After finding a trace or correlation id, use that value in a second narrow search instead of repeatedly scanning a broad time range. Use _source to keep only fields needed for the current investigation.
The tool rejects wildcard-only index segments such as *, **, or _all (including inside comma-separated index expressions), an empty search body, and body.size above 100. String size values are coerced and validated as numbers. The endpoint is an internal Kibana API, so compatibility should be verified after major Kibana upgrades.
📖 Resources
| Resource URI | Description |
|---|---|
kibana-api://paths |
List all available API endpoints |
kibana-api://paths?search=<keyword> |
Search endpoints by keyword |
kibana-api://path/{method}/{encoded_path} |
Get specific endpoint details |
💬 Example Queries
Basic Operations
- "What is the status of my Kibana server?"
- "List all available Kibana spaces"
- "Show me all API endpoints related to dashboards"
Elasticsearch & Logs
- "Search the production logs for errors from service X in the last hour"
- "Find this trace id and show the surrounding log events in chronological order"
- "Search index application-logs-* using this Elasticsearch Query DSL"
- "Aggregate error counts by service for this time range"
Saved Objects
- "Search for all dashboards"
- "Find visualizations containing 'nginx' in the title"
- "Create a new dashboard named 'Sales Overview'"
- "Update the description of dashboard 'my-dashboard-123'"
- "Delete multiple dashboards by their IDs"
Health & Analysis
- "Check health of dashboard 'overview'"
- "Analyze dependencies for visualization 'viz-123'"
- "Scan all dashboards for health issues"
🐛 Troubleshooting
Connection Issues
- Verify Kibana URL is accessible
- Check authentication credentials
- For SSL issues:
NODE_TLS_REJECT_UNAUTHORIZED=0(use with caution)
Claude Desktop Issues
- Restart Claude Desktop after config changes
- Validate JSON config syntax
- Check console logs for errors
Common Errors
- "import: command not found": Update to latest version
- Authentication failed: Verify credentials and permissions
- SSL errors: Check CA certificate or disable SSL validation
- 403 from
/api/console/proxywhile searching Elasticsearch: Usesearch_elasticsearch. The Console proxy can require the Kibana Dev Tools privilege even when the user has permission to read the target indices.
🔍 Debugging
Use MCP Inspector for debugging:
npm run inspector
This provides a browser-accessible debugging interface.
📦 Package Information
- NPM: @tocharianou/mcp-server-kibana
- GitHub: TocharianOU/mcp-server-kibana
- Node.js: >= 18.0.0
- License: Apache 2.0
🤝 Contributing
This project is community-maintained. Contributions and feedback are welcome!
Please follow the Elastic Community Code of Conduct in all communications.
📄 License
Apache License 2.0 - See LICENSE file for details.
Facts
- Kind
- MCP server
- Group
- Uncategorized
- Stars
- 84
- License
- Apache-2.0
- Language
- TypeScript
- Last push
- 2026-10-11
- Forks
- 28
- MCP Registry
- io.github.TocharianOU/kibana
- 1REAio.github.morluto/reaReverse engineer anything from your terminal or agent with one CLI and MCP server.125.3k
- 2com.browser-use/browser-usecom.browser-use/browser-useControl a real Chrome browser to complete any task: fill forms, extract data, book flights.117.7k
- 3World Monitorapp.worldmonitor/mcpLive markets, conflicts, country risk, chokepoints, energy, and China decision signals. 93 tools.88.3k
- 4Scrapling MCP Serverd4vinci/scrapling🕷️ An adaptive Web Scraping framework that handles everything from a single request to a full-scale crawl! Don't be shy, join here: https://discord.gg/EMgGbDceNQ and follow here for daily tips and tricks: https://x.com/Scraplingdev86.9k
- 5Everythingmodelcontextprotocol/serversThis MCP server attempts to exercise all the features of the MCP protocol. It is not intended to be a useful server, but rather a test server for builders of MCP clients. It implements prompts, tools, resources, sampling, and more to showcase MCP capabilities.85.8k
- 6Fetchmodelcontextprotocol/serversA Model Context Protocol server that provides web content fetching capabilities. This server enables LLMs to retrieve and process content from web pages, converting HTML to markdown for easier consumption.85.8k