Zabbix MCP Server
zabbix-mcp-server
An MCP server for the whole Zabbix API: 237 tools, several Zabbix servers, scoped tokens and a read-only mode
- 229 ★ on the repository
- Last commit
- AGPL-3.0
Broad access
Why this level
- The full Zabbix API: editing and deleting hosts, templates and triggers
- The Zabbix token opens monitoring of the whole infrastructure
- The read-only mode has to be turned on by you
Install
- Works in
Manual install
sudo ./deploy/install.shInstalls as a systemd service from a repository clone. Without root: ./deploy/install-user.sh. The Zabbix address and token are set in /etc/zabbix-mcp/config.toml.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The full Zabbix API: editing and deleting hosts, templates and triggers
- The Zabbix token opens monitoring of the whole infrastructure
- The read-only mode has to be turned on by you
This is third-party code. Review the repository files before installing.
What it does
The server covers the full Zabbix API. The README lists 237 tools, including 14 consolidated ones: host status, active problems, an infrastructure summary and the generic zabbix_raw_api_call. Several Zabbix servers can be connected. Client access is protected by OAuth 2.1 and scoped tokens with an IP restriction and an expiry. A read-only mode is set at the server and token level. There is an admin web portal on port 9090: tokens, users, templates and audit, and PDF reports in beta. The server runs as its own service and accepts clients over HTTP, SSE or stdio.
Who it is for. For monitoring administrators and DevOps engineers who work with Zabbix and want to ask an agent about hosts, problems and templates.
Good fit when
- You need a summary of active problems and host status from a plain question
- You need to prepare an audit of templates and triggers
- You need shared team access to Zabbix with tokens and a call audit
Not a fit when
- You need a simple setup without your own service: the server installs by script or Docker and needs configuration
- You are not ready to give an agent rights to change hosts, templates and triggers
Example request
Show active problems on all hosts for the last day, group them by severity and name the three noisiest triggersMCP
- Transport
- http, sse, stdio
- Authentication
- API key
Limitations
The full Zabbix API means editing and deleting hosts, templates and triggers. The read-only mode reduces the danger, but you have to turn it on yourself at the server and token level. A Zabbix API token is required: versions 5.0 and 5.2 have no API tokens. Python 3.10 or newer is needed. There is no PyPI package: the server installs from sources by the script `sudo ./deploy/install.sh` (systemd), by `./deploy/install-user.sh` without root or through Docker, and the Zabbix address and token are set in `/etc/zabbix-mcp/config.toml`. License: AGPL-3.0.
How to disable. Stop the zabbix-mcp service, remove the server entry from the client configs and revoke the Zabbix API token in the Zabbix interface.
FAQ
Is there a read-only mode?
Yes, it is set at the server and token level. You have to turn it on yourself: by default the server can change data.
How do I install it?
There is no PyPI package. It installs with the deploy/install.sh script with systemd, the install-user.sh script without root or in Docker.
What do I need from Zabbix?
A Zabbix API token. Versions 5.0 and 5.2 have no API tokens.
How many tools are there?
237 per the README, 14 of them consolidated, for example host status and active problems.
Discussion
No questions or reviews yet.
Related
MCP serversAn MCP server for Prometheus: the agent runs PromQL queries, discovers metrics and checks scrape targets
A self-hosted gateway that merges several MCP servers into one endpoint with auth and tool selection
Grafana MCP Server
Grafana MCP server
Grafana's official MCP server: dashboards, Prometheus and Loki queries, alerts and incidents straight from the agent
Official Cloudflare MCP server: the agent drives the whole Cloudflare API by searching the spec and running code
Datadog's official managed MCP server: logs, metrics, traces and incidents right in your agent
A single-container web dashboard for monitoring and controlling a fleet of 1C MCP servers: status, indexing progress, VRAM and license checks
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Zabbix MCP Server (this entry) | no | API key | 229 | |
| Prometheus MCP Server | no | not required | 518 | PromQL queries and Prometheus metrics |
| Grafana MCP Server | yes | API key | 3,558 | Grafana dashboards and queries to Prometheus and Loki |
| Netdata MCP | no | not required | 80,912 | metrics, logs and alerts of the Netdata agent |
README in short
The README describes a server for the full Zabbix API: 237 tools, several Zabbix servers, OAuth 2.1 and scoped tokens, a read-only mode, a web portal on port 9090 and PDF reports in beta. Installation is by script or Docker, and there is no PyPI package. AGPL-3.0 license.