Log inSubmit server
Cloud Tools logo

Cloud Tools

Read-only MCP tools for multi-cloud cost, inventory, waste, commitments, and recommendations across AWS, GCP, Cloudflare, and OVH.

InstallRepositoryWebsite
Full README Compare alternatives Embed widget Claim this listing

Install Cloud Tools

Pick your client, then add it in one click or paste the config.

STDIONode.js
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "cloud-tools": {
      "command": "npx",
      "args": [
        "-y",
        "@munhq/cloud-tools"
      ],
      "env": {
        "AWS_ACCESS_KEY_ID": "YOUR_VALUE_HERE",
        "AWS_SECRET_ACCESS_KEY": "YOUR_VALUE_HERE",
        "AWS_PROFILE": "YOUR_VALUE_HERE",
        "GOOGLE_APPLICATION_CREDENTIALS": "YOUR_VALUE_HERE",
        "CLOUDFLARE_API_TOKEN": "YOUR_VALUE_HERE",
        "CLOUDFLARE_ACCOUNT_ID": "YOUR_VALUE_HERE",
        "OVH_APPLICATION_KEY": "YOUR_VALUE_HERE",
        "OVH_CONSUMER_KEY": "YOUR_VALUE_HERE"
      }
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Values are written into the snippet above so you can paste it as is. They never leave your browser.

Automated check passed— started and listed 7 tools correctly (24d ago).

Overview

The cloud-tools MCP server exposes read-only cost, inventory, waste, commitment, and recommendation analysis for AWS, GCP, Cloudflare, and OVH. It runs locally, reads provider credentials from its environment, and exposes MCP over stdio or streamable HTTP, with an optional HTTP API. Use it when an agent needs to inspect cloud spend or resource waste without placing credentials in tool calls.

Use cases

  • Check access to multiple cloud accounts before analysis
  • Compare AWS and GCP costs using billing exports
  • Identify idle and oversized AWS or GCP resources
  • Review inventory across GCP, Cloudflare, and OVH
  • Summarize reachable cloud providers for an AI agent

Key features

  • Read-only multi-cloud cost analysis
  • AWS and GCP waste detection
  • Cross-cloud access and summary checks
  • Provider-specific inventory queries
  • MCP stdio and streamable HTTP transports
  • Optional HTTP API

Tools7

What Cloud Tools lets your agent do. Expand a tool for its parameters.

Verified live~5.7k tokens of context

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked 2d ago

Recent health check succeeded.

7/7 checks healthy over the last 40d

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 13d ago via OSV.dev · @munhq/cloud-tools (npm)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

npm downloads132Package downloads in the last 30 days.
Last commit24d agoMost recent push to the default branch.
Availability100%Our rolling endpoint + install checks that succeeded.
Install checkPassedOur sandbox started it and listed its tools.
Tools exposed7Callable tools this server registers over MCP.
Directory activity3 viewsConfig copies, upvotes, and views on AllMCPs.

How Cloud Tools works

What cloud-tools does

The cloud-tools MCP server gives an AI agent read-only access to selected cost, inventory, waste, commitment, and recommendation data across AWS, GCP, Cloudflare, and OVH. It is listed in the Cloud Platforms category and can run locally as a compiled Rust binary.

The tool set covers cloud access checks, cost reporting, cost comparison, inventory queries, waste analysis, commitment analysis, recommendations, and cross-cloud summaries. Provider coverage varies by operation. AWS and GCP support waste analysis; GCP, Cloudflare, and OVH support inventory; and GCP recommendations are available through get_recommendations. Unsupported operations return an explicit message rather than an empty result.

How it works

An agent supplies a cloud value such as aws or gcp to the relevant MCP tool. Credentials stay in the environment of the machine running cloud-tools and are not passed through tool arguments. The server contacts the configured provider APIs and returns results to the MCP client.

check_access is the recommended first call. It does not contact a cloud provider and reports which provider configurations are available and what is missing. get_cross_cloud_summary aggregates the clouds the server can reach.

Waste and inventory responses include coverage information and an errors list. A zero result with complete coverage indicates that no matching waste was found; a partial result indicates that some API calls failed and should not be treated as a clean account review.

Setup and configuration

The cloud-tools MCP server can be started with the npx wrapper, a release binary, a Docker container, or a source build. The documented MCP command is:

Terminal
claude mcp add cloud-tools -- npx -y @munhq/cloud-tools

The README also provides JSON configuration for Claude Desktop, Cursor, Windsurf, Zed, and Cline, plus installation links for Cursor and VS Code. The server supports stdio and streamable HTTP MCP transports. It also includes an optional Axum-based HTTP API.

AWS credentials can come from access-key variables, the shared AWS credentials file, an ECS task role, or an EC2 instance role. AWS role assumption can be configured with a role ARN and optional external ID. GCP uses Application Default Credentials or a service-account file, with optional project and BigQuery billing-table settings. Cloudflare requires a read-access API token and account ID. OVH uses application, secret, and consumer keys plus an optional endpoint.

The npm package is not released at the README's stated v0.1.0 status; until the next release is tagged, build from source with cargo build --release or use a release binary when available.

Tools and capabilities

The cloud-tools MCP server provides these documented operations:

  • check_access for provider configuration checks
  • get_costs for supported provider spend data
  • compare_costs for AWS and GCP comparisons, with a GCP billing table required
  • get_inventory for GCP, Cloudflare, and OVH resources
  • get_waste for AWS and GCP waste analysis
  • get_commitments for AWS and GCP commitments
  • get_recommendations for GCP recommendations
  • get_cross_cloud_summary for reachable-cloud summaries

AWS waste analysis examines resources including EC2, RDS, S3, Lambda, DynamoDB, ECS, ElastiCache, load balancers, NAT gateways, CloudWatch log groups, and Compute Optimizer data. Idle and oversized findings use CloudWatch CPU series.

Limitations and notes

GCP cost reporting depends on a BigQuery billing export for actual spend. Without one, get_costs falls back to budget amounts and labels the source accordingly; compare_costs requires the billing table. AWS does not yet provide an inventory tool. Cloudflare and OVH do not provide waste analysis.

The server is documented as collecting no telemetry or analytics. Provider credentials are held in memory for the duration of a call, and the npm wrapper caches only the downloaded binary under ~/.cache/cloud-tools/bin/.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about Cloud Tools

cloud-tools is an MCP server that connects AI agents to AWS, GCP, Cloudflare, and OVH. Its main tools check access, retrieve costs and inventory, analyze waste, inspect commitments, provide GCP recommendations, compare supported costs, and create cross-cloud summaries. It operates locally and keeps provider credentials out of tool calls.

Install it with the documented `claude mcp add cloud-tools -- npx -y @munhq/cloud-tools` command, or add its stdio configuration to an MCP client.

cloud-tools supports AWS, GCP, Cloudflare, and OVH. Available operations differ by provider: AWS and GCP support waste analysis, GCP, Cloudflare, and OVH support inventory, and GCP supports recommendations. The access check reports which configured providers are reachable.

Yes, when a BigQuery billing export is configured through `CLOUD_TOOLS_GCP_BILLING_TABLE`. Without that table, `get_costs` falls back to the Budgets API and identifies the result as budget amounts rather than actual spend. GCP cost comparison requires the billing table.

Yes, cloud-tools has a Cursor installation link and documented JSON configuration using the `npx` command. Cursor runs the server through MCP stdio, while provider credentials remain in the environment of the machine running cloud-tools.

Related MCP servers

All in Cloud Platforms All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewCloud Tools AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/cloud-tools?style=directory)](https://allmcps.com/mcp/cloud-tools)
HTML Embed
<a href="https://allmcps.com/mcp/cloud-tools"><img src="https://allmcps.com/api/badge/cloud-tools?style=directory" alt="Cloud Tools on AllMCPs" /></a>