# cloudpulse [Health: Active]

**Category:** ☁️ Cloud Platforms  
**Repository:** https://github.com/Galadriel-Tech-Solutions/cloudpulse-mcp  
**GitHub Stars:** 1  
**npm Downloads (last month):** 145  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/cloudpulse

## Description
Cross-cloud observability for AI agents. Covers AWS, GCP, Vercel, and Cloudflare.

## Tools
Capabilities this server exposes over MCP:

- **list_cloud_topology** — Scan all configured platforms and return a unified service map.
- **get_correlated_logs** — Fetch and merge logs from Vercel + AWS CloudWatch into one timeline.
- **diagnose_service_link** — Check why service A can't reach resource B.
- **check_resource_limits** — Query quotas and flag resources nearing their limits.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "cloudpulse": {
    "command": "npx",
    "args": ["-y","cloudpulse-mcp"],
    "env": {
      "AWS_ACCESS_KEY_ID": "",
      "AWS_SECRET_ACCESS_KEY": "",
      "AWS_PROFILE": "",
      "AWS_REGION": "",
      "VERCEL_TOKEN": "",
      "VERCEL_TEAM_ID": "",
      "GOOGLE_APPLICATION_CREDENTIALS": "",
      "CLOUDFLARE_API_TOKEN": ""
    }
  }
}
```

**Requires environment variables:** `AWS_ACCESS_KEY_ID`, `AWS_SECRET_ACCESS_KEY`, `AWS_PROFILE`, `AWS_REGION`, `VERCEL_TOKEN`, `VERCEL_TEAM_ID`, `GOOGLE_APPLICATION_CREDENTIALS`, `CLOUDFLARE_API_TOKEN` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What the cloudpulse MCP server does

The cloudpulse MCP server connects an MCP-compatible AI client to selected AWS, Vercel, GCP, and Cloudflare APIs. Its purpose is operational inspection rather than resource management: the tools read service metadata, logs, network-related configuration, and quota information without storing credentials or changing infrastructure.

The server is useful when an application spans providers and the relevant evidence is split between consoles. For example, an agent can inspect a Vercel deployment alongside AWS resources, or review Cloud Run, Cloud SQL, and Cloud Logging data when those credentials are configured.

## How it works

Cloudpulse discovers which providers are configured and uses their APIs when a tool is called. AWS access can come from access-key environment variables, an AWS profile, or an EC2 instance role. Other providers use their documented token or application-credential environment variables. Values are read at call time, and the project states that credentials are neither logged nor persisted.

The cloudpulse MCP server communicates over MCP and exposes four tools. Inputs can narrow scans by platform, region, project, time range, trace ID, or other provider-specific values. Some fields are required for a particular operation, while optional fields use defaults such as the current time, the `/aws/lambda` log-group prefix, or an 80 percent warning threshold.

## Setup and configuration

The quickest launch method is:

```bash
npx cloudpulse-mcp
```

The package requires Node.js 18 or newer. Claude Desktop and Cursor can start it with an MCP configuration that invokes `npx -y cloudpulse-mcp`. VS Code 1.99 or newer with GitHub Copilot in Agent Mode uses a built project and starts `dist/index.js` as a stdio server. A source checkout can be installed with `npm install`, run during development with `npm run dev`, and compiled with `npm run build`.

Configure only the providers needed for the investigation. AWS supports `AWS_ACCESS_KEY_ID` and `AWS_SECRET_ACCESS_KEY`, `AWS_PROFILE`, or an instance role. Vercel uses `VERCEL_TOKEN`, with `VERCEL_TEAM_ID` available for team access. GCP uses `GOOGLE_APPLICATION_CREDENTIALS`. Cloudflare uses `CLOUDFLARE_API_TOKEN` and `CLOUDFLARE_ACCOUNT_ID`.

## Tools and capabilities

- `list_cloud_topology` returns a unified map of active services across configured platforms. Platform filters and an AWS region can be supplied.
- `get_correlated_logs` merges Vercel and AWS CloudWatch events into one timeline. It accepts a required start time, an optional end time, trace or request ID, project, log-group prefix, and region.
- `diagnose_service_link` investigates why a source service cannot reach a target. It checks relevant Vercel database environment variables, AWS security-group ingress, and external API reachability; ports can be supplied or inferred for common targets.
- `check_resource_limits` queries quotas and flags usage at a configurable threshold, defaulting to 80 percent.

## Limitations and notes

The current implementation is read-only and does not provide deployment, mutation, or remediation operations. Cross-provider log correlation specifically covers Vercel and AWS CloudWatch, while topology and quota coverage depends on configured credentials and supported provider data. The documented roadmap lists diagnostic playbooks for CORS, 504 timeouts, and cold-start loops as future work, so those workflows should not be assumed to be built in. The project is licensed under MIT.

_Full upstream README: https://allmcps.com/mcp/cloudpulse/readme_