The observability hub for the AI-agent era

Cursor, Claude Code, and Codex can now invoke idcd's global probe capabilities through the standard MCP protocol.

Typical use cases

What your AI assistant can do now

Have Claude / Cursor run Ping, MTR, and SSL expiry checks from global nodes directly. The AI produces a root-cause report without anyone SSHing in.

“Check api.example.com latency from Japan and Europe, give me an MTR route map.”

Wire a Service Token into your pipeline; every deploy auto-calls probe_http + probe_ssl to verify rollout health. AI summarizes when something fails.

“After deploy, verify the HTTPS cert and API latency; alert if p95 > 800ms.”

When your LLM app calls external APIs, idcd MCP measures each outbound link in real time. Agents flag slow APIs and suggest alternatives.

“Compare OpenAI vs Claude API latency from mainland China — recommend the best endpoint.”

.cursor/mcp.json

{
  "mcpServers": {
    "idcd": {
      "url": "https://mcp.idcd.com/sse",
      "token": "YOUR_TOKEN"
    }
  }
}

terminal

claude mcp add idcd \
  --transport sse \
  https://mcp.idcd.com/sse

terminal

export MCP_SERVER_URL=https://mcp.idcd.com/sse
export MCP_TOKEN=YOUR_TOKEN

codex --mcp $MCP_SERVER_URL

Full-stack network probing — agents always know the network state.

pingMeasure latency and packet loss to a host

httpHTTP/HTTPS availability and response time

dnsDNS resolution lookup

tracerouteTrace the network path to a host

sslSSL/TLS certificate validity and expiry

diagnoseOne-shot diagnostic (ping/DNS/HTTP/SSL)

ipIP geolocation and ASN lookup

whoisWHOIS registration for a domain or IP

Debug

24h

For personal debugging. Try out every MCP tool quickly.

Team

90d

Shared across a team — multiple members hit the same MCP endpoint.

Automation

90d, auto-renewing

For CI/CD pipelines. Transparent auto-refresh.

MCP access is included in every plan — try it on the Free tier.