SuperAPI AI Directory
Partner with us中文AI newsFavorites
All toolsAI agentsdatadog-mcp-server

datadog-mcp-server

Datadog observability: metrics, monitors, logs, APM, RUM, incidents, status pages, fleet automation and workflows, plus incident-triage prompts.

Open website

About this tool

Datadog MCP Server is an open-source Model Context Protocol (MCP) server designed to enable AI assistants to interact directly with the Datadog observability platform. It is community-maintained (us-all/datadog-mcp-server), built on the @us-all MCP standard, and provides 163 tools covering Datadog's 16 main API categories.

Purpose

The server's main purpose is to enable AI agents (such as Claude, ChatGPT) to query and operate monitoring data in Datadog, answering questions such as "Why is this happening?" rather than just "What is the value?" It is particularly suitable for scenarios requiring deep observability analysis, incident triage, SLO compliance checks, monitoring noise audits, and more.

Main Capabilities

  • Broad API coverage: 163 tools divided into 16 categories: metrics, monitors, dashboards, logs, apm, rum, incidents, security, synthetics, ci, infra, fleet, status-pages, oncall, teams, account.
  • Aggregation tools: such as analyze-monitor-state and slo-compliance-snapshot can collapse 5–7 sequential API calls into a single structured response, and include a caveats array to handle partial failures.
  • Full SLO CRUD: supports creating, updating, and deleting SLOs and their corrections, whereas the official Bits AI MCP and community alternatives are usually read-only.
  • Fleet Automation: 15 tools covering agents, deployments, and scheduling, provided only by this server.
  • Status Pages: 21 tools supporting the full lifecycle of status pages (components, degradations, maintenance).
  • Token efficiency design: through extractFields projections,DD_TOOLS / DD_DISABLE 16-category toggles, and the search-tools meta-tool, it keeps LLM context consumption low across 163 tools.
  • Apps SDK cards:slo-compliance-snapshot can be rendered as visual cards in the ChatGPT client via _meta["openai/outputTemplate"] ; the Claude client receives the same JSON content.
  • Transport methods: default stdio (for Claude Desktop / Code). Setting MCP_TRANSPORT=http enables Streamable HTTP, for the ChatGPT Apps SDK or remote clients (with Bearer authentication via MCP_HTTP_TOKEN ).

Usage

  1. Installation and configuration: The server can be added to ~/Library/Application Support/Claude/claude_desktop_config.json, or built from source. Specific configuration examples are provided (see official documentation).
  2. Connection: Connect the server to Claude Desktop or Claude Code, then natural language prompts can be used.
  3. Example prompts:
    • SLO health: "List my SLOs and their remaining error budget for this month. Group by status: compliant, at risk, breached."
    • Incident triage: "There is an active incident on checkout-service. Pull the associated monitors, recent APM error spikes, and deployments involving this service over the past 24 hours."
    • Monitoring noise audit: "Find monitors that alerted more than 10 times in the past 7 days but have an MTTR under 5 minutes—these may be flapping."
    • RUM error spike: "The RUM error rate for the checkout funnel jumped between 14:00 and 14:30 today. Show the top error groups, affected sessions, and the user actions before the error."
    • Fleet deployment: "Schedule datadog-agent 7.55.0 to start next Saturday, deploying only to the staging cluster on weekends."

Suitable Scenarios

  • AI agent workflows that require deep interaction with Datadog.
  • Self-hosted scenarios (internal sites, isolated networks, development/CI sandboxes).
  • Operations that require write parity for SLOs, Fleet, and Status Pages.
  • Teams that want to reduce LLM context consumption.

Limitations

  • Read-only by default; all write operations require setting DD_ALLOW_WRITE=true.
  • Some tools (such as Fleet, Status Pages) are provided only by this server and may be incompatible with the official MCP.
  • Requires a Node.js 22+ environment.
  • The official MCP (Bits AI MCP) is complementary rather than a replacement; the official one is suitable for quick hosted onboarding and SSO.

MCP Prompts and Resources

  • Provides 4 prompt templates:triage-incident、audit-monitor-noise、analyze-rum-error-spike、investigate-slow-trace.
  • Read-only entity URI:dd://monitor/{id}、dd://dashboard/{id}、dd://slo/{id} , etc.

Tech Stack

Node.js 22+, TypeScript strict ESM, pnpm, @modelcontextprotocol/sdk, @datadog/datadog-api-client (official), zod, dotenv, vitest + dd-trace. Based on @us-all/mcp-toolkit, providing shared patterns such as extractFields, aggregate, and createWrapToolHandler.

Related tools

Browse category