Connect your AI agents

Features

• Multi-file Composer
• Background agents
• Codebase indexing
• Terminal integration

1. 1

   Download Cursor

   Download and install Cursor from the official website.

   https://cursor.sh
2. 2

   Add Datadog extension

   Install the Datadog extension—MCP server access is included automatically.

   cursor --install-extension datadog.datadog-vscode
3. 3

   Open Composer

   Open Cursor's Composer (Cmd+I on Mac, Ctrl+I on Windows) for multi-file AI edits and start using Datadog MCP tools.

Example prompts

• "Add datadog tracing to all handlers in this file"

• "Build a trace analyzer that isolates error span chains"

• "Review which synthetic test failures are false alarms"

Features

• Full codebase understanding
• Terminal command execution
• Git operations
• Multi-file editing

1. 1

   Install Claude Code

   Install the Claude Code CLI globally.

   npm install -g @anthropic-ai/claude-code
2. 2

   Add Datadog MCP Server

   Register the Datadog MCP server with Claude Code (see full setup guide for other regions).

   claude mcp add --transport http datadog-mcp https://mcp.datadoghq.com/api/unstable/mcp-server/mcp
3. 3

   Start coding

   Launch Claude Code and use Datadog tools directly in your terminal.

   claude

Example prompts

• "Triage this monitor alert using logs, traces, and deploys"

• "Find idle services with only health check traffic"

• "Show me error rate for the checkout service this week"

Features

• 1M token context
• Shell execution
• MCP support
• Google Cloud integration

1. 1

   Install Gemini CLI

   Install the Gemini CLI from Google.

   npm install -g @google/gemini-cli
2. 2

   Add to Gemini settings

   Add to ~/.gemini/settings.json (see full setup guide for other regions).

   config

   {
     "mcpServers": {
       "datadog": {
         "type": "http",
         "url": "https://mcp.datadoghq.com/api/unstable/mcp-server/mcp"
       }
     }
   }

3. 3

   Start using

   Run Gemini CLI and query Datadog directly.

   gemini

Example prompts

• "Analyze error trends across all services"

• "Check monitor status for production"

• "Find the root cause of this latency spike"

Features

• Agent mode
• Datadog extension
• MCP tool access
• Multi-org switching

1. 1

   Install Datadog Extension

   Install the Datadog extension—MCP server access is included automatically.

   code --install-extension datadog.datadog-vscode
2. 2

   Sign in and restart

   Sign in to your Datadog account from the extension, then restart VS Code to activate MCP.

3. 3

   Enable Copilot Agent Mode

   Open Copilot chat and switch to Agent mode to use Datadog MCP tools. Requires an active GitHub Copilot subscription.

Example prompts

• "Use Datadog to check error rates"

• "Find traces for slow API calls"

• "List active monitors"

Features

• AI agent mode
• MCP native
• Shell context awareness
• Team collaboration

1. 1

   Download Warp

   Install Warp from the official website.

   https://warp.dev
2. 2

   Add Datadog MCP server

   In Warp Settings → MCP Servers, add a new server with URL: https://mcp.datadoghq.com/api/unstable/mcp-server/mcp. Authenticate via OAuth when prompted.

   OPEN WARP MCP SETTINGS
3. 3

   Use Agent mode

   Open Warp's agent mode (Ctrl+Shift+A) and start querying Datadog.

Example prompts

• "What are the main query error logs for my team's service in last 30min"

• "Get kubernetes pods with zero non-probe traffic today"

• "Compare p95 latency between deploy tags v2.2 and v2.3"

Features

• Autonomous coding
• End-to-end execution
• MCP integration
• CI/CD aware

1. 1

   Access Devin

   Sign up for Devin at Cognition's website.

   https://devin.ai
2. 2

   Add Datadog MCP

   In Devin's MCP marketplace, find Datadog and connect with URL: https://mcp.datadoghq.com/api/unstable/mcp-server/mcp. Authenticate via OAuth when prompted.

   ADD DATADOG TO DEVIN
3. 3

   Assign a task

   Give Devin a task and it will use Datadog tools to investigate and fix issues autonomously.

Example prompts

• "Investigate why the menu page is showing latency"

• "Create a cost optimization notebook scanning our dev AWS accounts and cost metrics"

• "Instrument this entire repo with Datadog RUM"

Features

• File read/write
• Command execution
• HTTP transport
• Sandboxed environment

1. 1

   Install Codex CLI

   Install the OpenAI Codex CLI.

   npm install -g @openai/codex
2. 2

   Add to Codex config

   Add to ~/.codex/config.toml (see full setup guide for other regions).

   config

   [mcp_servers.datadog]
   url = "https://mcp.datadoghq.com/api/unstable/mcp-server/mcp"

3. 3

   Authenticate

   Log in via OAuth to connect your Datadog account.

   codex mcp login datadog
4. 4

   Start using

   Run Codex and ask it to query your Datadog data.

   codex

Example prompts

• "What service is causing OOMKilled issues for my Kubernetes pods?"

• "Scan Datadog for services with no real user traffic"

• "Graph p50 and p99 latency for all API endpoints"

Features

• Remote MCP via HTTP
• Custom connector support
• Cross-platform
• Persistent sessions

1. 1

   Download Claude Desktop

   Install Claude Desktop from Anthropic's website.

   https://claude.ai/download
2. 2

   Add Datadog as a custom connector

   In Claude Desktop Settings → Integrations, add a custom connector with URL: https://mcp.datadoghq.com/api/unstable/mcp-server/mcp. Authenticate via OAuth when prompted.

   ADD DATADOG TO CLAUDE DESKTOP
3. 3

   Start using Datadog tools

   Open a conversation in Claude Desktop and start using Datadog MCP tools.

Example prompts

• "What's the current state of my infrastructure?"

• "Show me error trends over the last 24 hours"

• "Which services have the highest latency?"

Features

• Open source
• MCP native
• Extensible plugins
• Multi-provider

1. 1

   Install Goose

   Install the Goose CLI from Block.

   brew install block/tap/goose
2. 2

   Add Datadog MCP extension

   Run `goose configure`, select Extensions → Add, choose type Remote MCP, and enter URL: https://mcp.datadoghq.com/api/unstable/mcp-server/mcp. Authenticate via OAuth when prompted.

3. 3

   Start a session

   Launch Goose and begin using Datadog tools.

   goose session

Example prompts

• "What service is causing OOMKilled issues for my Kubernetes pods?"

• "Scan Datadog for services with no real user traffic"

• "Graph p50 and p99 latency for all API endpoints"

Features

• Open source
• Terminal UI
• MCP native
• Multi-provider support

1. 1

   Install OpenCode

   Install the OpenCode CLI.

   curl -fsSL https://opencode.ai/install | bash
2. 2

   Add to OpenCode config

   Add to your opencode.json config (see full setup guide for other regions).

   config

   {
     "mcp": {
       "datadog": {
         "type": "remote",
         "url": "https://mcp.datadoghq.com/api/unstable/mcp-server/mcp"
       }
     }
   }

3. 3

   Start coding

   Launch OpenCode and use Datadog tools directly.

   opencode

Example prompts

• "Correlate this monitor alert with recent deploys and feature flag changes"

• "What's the p99 latency for this API route today?"

• "Find the bottleneck in this 3000-span trace"

Features

• Agent mode
• Multi-file edits
• Terminal execution
• MCP tool use

1. 1

   Install Datadog Extension

   Install the Datadog extension for VS Code—MCP server access is included automatically.

   code --install-extension datadog.datadog-vscode
2. 2

   Sign in and restart

   Sign in to your Datadog account from the extension, then restart VS Code to activate MCP.

3. 3

   Use Agent mode

   Open Copilot chat, switch to Agent mode, and start using Datadog tools. Requires an active GitHub Copilot subscription.

Example prompts

• "Add Database Monitoring to the new postgres database we just added"

• "Analyze what is causing slow loads for the processing service"

• "What's the p99 latency for this API route today?"

Features

• Pre-built Datadog MCP integration
• Choose any LLM provider
• Add custom tools & prompts
• Full observability built-in

1. 1

   Pick your LLM

   Choose any model provider — OpenAI, Anthropic, or open-source.

2. 2

   Connect Datadog MCP

   Add the MCP server to your agent framework (see full setup guide for other regions).

   config

   {
     "mcpServers": {
       "datadog": {
         "type": "http",
         "url": "https://mcp.datadoghq.com/api/unstable/mcp-server/mcp"
       }
     }
   }

3. 3

   Add custom tools

   Extend your agent with custom tools, prompts, and workflows.

4. 4

   Deploy & observe

   Ship your agent and monitor it with Datadog from day one.

Agent ideas

• "Deploy-Aware Incident Response Agent"

• "Auto-Fix Agent (From Trace to Pull Request)"

• "Observability Best Practices Onboarding Agent"

• "Zombie Service Hunter"

• "Feature Flag ↔ Incident Correlation Agent"

• "Cloud Cost Anomaly Agent"