feat: add signoz_list_hosts tool for infrastructure monitoring

[artgas1]

Summary

Adds a new signoz_list_hosts tool that exposes SigNoz's Infrastructure Monitoring hosts data via MCP. This enables AI assistants to check server health, CPU/memory usage, and load averages directly.

New Tool

signoz_list_hosts — List infrastructure hosts with resource utilization metrics.

Parameters

Parameter	Type	Default	Description
timeRange	string	1h	Time range (e.g., '30m', '1h', '6h', '24h')
orderBy	string	cpu	Sort column: cpu, memory, wait, load15
order	string	desc	Sort direction: asc, desc
limit	string	100	Max hosts to return
offset	string	0	Pagination offset

Example Response

{
  "hosts": [
    {
      "hostName": "production-vps",
      "active": true,
      "os": "linux",
      "cpu": 0.33,
      "memory": 0.36,
      "wait": 0.004,
      "load15": 2.05
    }
  ],
  "total": 5,
  "sentAnyHostMetricsData": true
}

Implementation

• Client: ListHosts() → POST /api/v1/hosts/list
• Types: HostListRequest, HostListResponse, HostRecord in new pkg/types/infra.go
• Handler: Parses time range, sorting, pagination; simplifies response by extracting OS from metadata
• Server: Registered via RegisterInfraHandlers()

Use Case

This was motivated by a real incident where a VPS became overloaded but we couldn't query host metrics through MCP. The existing signoz_execute_builder_query requires complex query construction for host metrics, while this tool provides a simple, purpose-built interface.