Mcp Vmanomaly
Expose VictoriaMetrics vmanomaly detection APIs to your agent so you can investigate anomalies and tune models without manual HTTP calls.
Overview
vmanomaly MCP is an Operate-phase MCP server that lets coding agents interact with VictoriaMetrics vmanomaly for metrics anomaly detection APIs.
What is this MCP server?
- MCP server v0.2.3 for VictoriaMetrics vmanomaly anomaly detection service
- Configurable VMANOMALY_ENDPOINT (e.g. http://localhost:8490) with optional bearer token
- Custom headers via VMANOMALY_HEADERS comma-separated key=value pairs
- MCP_SERVER_MODE supports stdio (default), sse, and http transports
- SSE/HTTP modes use MCP_LISTEN_ADDR (default localhost:8080)
- Server version 0.2.3
- OCI identifier ghcr.io/victoriametrics/mcp-vmanomaly:0.2.3
- Required env VMANOMALY_ENDPOINT
Community signal: 8 GitHub stars.
What problem does it solve?
Manually hitting vmanomaly REST endpoints while debugging odd metric behavior fragments context between your terminal and your AI pair programmer.
Who is it for?
Solo builders operating VictoriaMetrics plus vmanomaly who want agent-assisted anomaly review and API exploration.
Skip if: Projects without vmanomaly deployed or teams that only need raw metric queries without anomaly scoring.
What do I get? / Deliverables
With VMANOMALY_ENDPOINT configured, your agent can drive vmanomaly workflows in chat while you refine alerts and monitoring during production iteration.
- Agent-facing vmanomaly API integration
- Flexible MCP transport (stdio, sse, or http per MCP_SERVER_MODE)
- Streamlined anomaly investigation during monitoring work
Recommended MCP Servers
Journey fit
Anomaly detection on production metrics is an Operate concern once the product is live and instrumented. Monitoring is the right shelf because vmanomaly augments alerting and SRE workflows, not greenfield backend coding.
How it compares
Anomaly-service MCP bridge, not the vmanomaly engine installer or a general-purpose ML notebook.
Common Questions / FAQ
Who is mcp-vmanomaly for?
Operators and developers who run vmanomaly alongside VictoriaMetrics and want MCP tools for anomaly-related API work from their coding agent.
When should I use mcp-vmanomaly?
Use it in Operate monitoring when investigating automated anomaly alerts, validating detector behavior, or iterating on observability after launch.
How do I add mcp-vmanomaly to my agent?
Register the stdio MCP package (ghcr.io/victoriametrics/mcp-vmanomaly:0.2.3), set VMANOMALY_ENDPOINT, optional VMANOMALY_BEARER_TOKEN and headers, and choose MCP_SERVER_MODE if you need sse or http.