# Amazon CloudWatch

Amazon CloudWatch connects AI agents to CloudWatch metrics, alarms, and log groups. It speeds up root cause diagnostics across AWS infrastructure.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · ★ 9939 repo stars
- Author: awslabs
- License: Apache-2.0 · policy=link-only · description only; the file stays with the author
- Source: https://github.com/awslabs/mcp/tree/HEAD/src/cloudwatch-mcp-server
- Updated: 2026-10-23
- Security check: automated, 2026-11-16 (text) — no signs of malicious code found.
- HTML: https://affpapa.org/en/nevibecode/mcp/amazon-cloudwatch

## What it does

This MCP server exposes AWS operational telemetry directly to troubleshooting agents. It maps CloudWatch endpoints into dedicated tools for metrics retrieval, PromQL execution, alarm analysis, and log searches.

For performance tracking, `get_metric_data` handles percentiles such as p90 or p99, math expressions, and multi-metric batching. Log analysis includes `execute_cwl_insights_batch`, which runs Logs Insights queries across multiple AWS regions simultaneously. That tool splits batches into chunks of 50 log groups, caps concurrency at 7 per region, and merges the results. To query enriched AWS vended metrics with PromQL, you must first enable OpenTelemetry via `aws cloudwatch start-otel-enrichment`.

You can only run this server locally on the host where your LLM client operates. It requires valid AWS credentials preconfigured in environment variables or via `aws configure`.

## Requirements
- Needs: Docker

## Install

### Claude Code (native platform)

**Install**
- The MCP registry has no launch command for it: see the repository README

## When to use it
- inspect active AWS alarms to pinpoint incident root causes
- query log groups across multiple regions in one batch operation
- extract latency percentiles and evaluate custom metric expressions
- explore time-series telemetry using PromQL selectors

## Example prompts
- Check active CloudWatch alarms and retrieve recent state transitions
- Run a Logs Insights query on `/aws/lambda/backend` for errors in the last hour
- Pull p99 latency metrics and recommend alarm thresholds for this service

## Similar
- [setting-up-cloudwatch-alarm-notifications](https://affpapa.org/en/nevibecode/skills/setting-up-cloudwatch-alarm-notifications.md) — skill · 7200 installs
- [setting-up-cloudwatch-observability](https://affpapa.org/en/nevibecode/skills/setting-up-cloudwatch-observability.md) — skill · 2100 installs
- [aws-cloudformation-cloudwatch](https://affpapa.org/en/nevibecode/skills/aws-cloudformation-cloudwatch.md) — skill · ★ 372 repo stars
- [aws-cloudwatch-investigation](https://affpapa.org/en/nevibecode/skills/aws-cloudwatch-investigation.md) — skill · 43 installs
- [dt-obs-aws](https://affpapa.org/en/nevibecode/skills/dt-obs-aws.md) — skill · 2800 installs
- [aws-cost-operations](https://affpapa.org/en/nevibecode/skills/aws-cost-operations.md) — skill · 921 installs

From the same set: [awslabs/mcp](https://affpapa.org/en/nevibecode/by/awslabs/mcp.md) — 36 cards

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md
