VictoriaMetrics Cloud exposes a hosted MCP (Model Context Protocol) server that allows MCP-compatible AI assistants to query and explore your metrics deployments using natural language.
This integration enables AI tools to access your VictoriaMetrics Cloud metrics directly, making it easier to investigate incidents, explore data, and build queries without leaving your AI assistant.
Only metrics are currently supported. Logs support is planned for a future release.
Integrating AI Assistants via MCP #
Unlike other VictoriaMetrics Cloud integrations, the hosted MCP server uses an API key for authentication rather than a per-deployment access token. You can configure the deployments this API key has access to.
You can manage them in the API keys page in the console, or read more in the dedicated documentation page .
Avoid sharing configuration files that contain your API key in plaintext. If your AI tool supports
environment variables in MCP configurations, use them instead of storing the key directly.
For example, ${VMCLOUD_API_KEY}.
Connecting your AI Assistant to the hosted MCP server #
To connect your AI assistant to VictoriaMetrics Cloud via MCP, visit the cloud console , or follow this interactive guide:
Running the MCP Server Locally #
If you prefer to run the MCP server on your own infrastructure, you can set it up locally by providing the bearer token for your deployment, its instance type, and entry point. See the setup instructions in the mcp-victoriametrics repository for details.