Skip to main content
Glama

GCP MCP

A Model Context Protocol (MCP) server that enables AI assistants like Claude to interact with your Google Cloud Platform environment. This allows for natural language querying and management of your GCP resources during conversations.

Features

  • 🔍 Query and modify GCP resources using natural language

  • ☁️ Support for multiple GCP projects

  • 🌐 Multi-region support

  • 🔐 Secure credential handling (no credentials are exposed to external services)

  • 🏃‍♂️ Local execution with your GCP credentials

  • 🔄 Automatic retries for improved reliability

Related MCP server: GCP Billing and Monitoring MCP Server

Prerequisites

  • Node.js

  • Amazon Q CLI/Claude Desktop/Cursor/Windsurf

  • GCP credentials configured locally (application default credentials)

Installation

  1. Clone the repository:

git clone https://github.com/RadiumGu/gcp-ops-mcp.git
cd gcp-ops-mcp
  1. Install dependencies:

npm install
  1. Build the project:

npm run build

Configuration

Amazon Q CLI

Add the following configuration to your Q CLI MCP settings:

{
  "mcpServers": {
    "gcp-mcp": {
      "command": "node",
      "args": ["/path/to/gcp-ops-mcp/bin.js"],
      "env": {
        "NODE_PATH": "/path/to/gcp-ops-mcp/node_modules"
      }
    }
  }
}

Replace /path/to/gcp-ops-mcp with the actual path to your project directory.

Claude Desktop

  1. Open Claude desktop app and go to Settings -> Developer -> Edit Config

  2. Add the following entry to your claude_desktop_config.json:

via npm:

{
  "mcpServers": {
    "gcp": {
      "command": "sh",
      "args": ["-c", "npx -y gcp-mcp"]
    }
  }
}

If you installed from source:

{
  "mcpServers": {
    "gcp": {
      "command": "npm",
      "args": [
        "--silent",
        "--prefix",
        "/path/to/gcp-mcp",
        "start"
      ]
    }
  }
}

Replace /path/to/gcp-mcp with the actual path to your project directory if using source installation.

Cursor

  1. Open Cursor and go to Settings (⌘,)

  2. Navigate to AI -> Model Context Protocol

  3. Add a new MCP configuration:

{
  "gcp": {
    "command": "npx -y gcp-mcp"
  }
}

Windsurf

  1. Open ~/.windsurf/config.json (create if it doesn't exist)

  2. Add the MCP configuration:

{
  "mcpServers": {
    "gcp": {
      "command": "npx -y gcp-mcp"
    }
  }
}

GCP Setup

  1. Set up GCP credentials:

    • Set up application default credentials using gcloud auth application-default login

  2. Refresh your AI assistant (Claude Desktop/Cursor/Windsurf)

Usage

Start by selecting a project or asking questions like:

  • "List all GCP projects I have access to"

  • "Show me all Cloud SQL instances in project X"

  • "What's my current billing status?"

  • "Show me the logs from my Cloud Run services"

  • "List all GKE clusters in us-central1"

  • "Show me all Cloud Storage buckets in project X"

  • "What Cloud Functions are deployed in us-central1?"

  • "List all Cloud Run services"

  • "Show me BigQuery datasets and tables"

Available Tools

  1. run-gcp-code: Execute GCP API calls using TypeScript code

  2. list-projects: List all accessible GCP projects

  3. select-project: Select a GCP project for subsequent operations

  4. get-billing-info: Get billing information for the current project

  5. get-cost-forecast: Get cost forecast for the current project

  6. get-billing-budget: Get billing budgets for the current project

  7. list-gke-clusters: List all GKE clusters in the current project

  8. list-sql-instances: List all Cloud SQL instances in the current project

  9. get-logs: Get Cloud Logging entries for the current project

Example Interactions

  1. List available projects:

List all GCP projects I have access to
  1. Select a project:

Use project my-project-id
  1. Check billing status:

What's my current billing status?
  1. View logs:

Show me the last 10 log entries from my project

Supported Services

  • Google Compute Engine

  • Cloud Storage

  • Cloud Functions

  • Cloud Run

  • BigQuery

  • Cloud SQL

  • Google Kubernetes Engine (GKE)

  • Cloud Logging

  • Cloud Billing

  • Resource Manager

  • More coming soon...

Troubleshooting

To see logs:

tail -n 50 -f ~/Library/Logs/Claude/mcp-server-gcp.log

Common issues:

  1. Authentication errors: Ensure you've run gcloud auth application-default login

  2. Permission errors: Check IAM roles for your account

  3. API errors: Verify that required APIs are enabled in your project

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

MIT

Available Tools

9 tools
get-billing-budgetC

Get billing budgets for the current project

Inputs: projectId.

get-billing-infoC

Get billing information for the current project

Inputs: projectId.

get-cost-forecastB

Get cost forecast for the current project

Inputs: projectId, months.

get-logsB

Get Cloud Logging entries for the current project

Inputs: filter, pageSize.

list-gke-clustersB

List all GKE clusters in the current project

Inputs: location.

list-projectsA

List all GCP projects accessible with current credentials

Inputs: no named parameters.

list-sql-instancesB

List all Cloud SQL instances in the current project

Inputs: no named parameters.

run-gcp-codeD

Run GCP code

Inputs: reasoning (required), code (required), projectId, region.

select-projectB

Selects GCP project to use for subsequent interactions

Inputs: projectId (required), region.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    C2.8/5.0

    Scored across 9 tools

    Disambiguation4/5

    Most tools have distinct purposes targeting specific GCP services or functions, but 'run-gcp-code' is ambiguous and could overlap with other operations, potentially causing confusion. The other tools are clearly scoped to billing, logging, resource listing, and project selection.

    Naming Consistency3/5

    The naming is mixed with inconsistent patterns: some use verb_noun (e.g., 'list-projects', 'get-logs'), others use noun_verb (e.g., 'run-gcp-code'), and hyphenation varies. While readable, it lacks a uniform convention across all tools.

    Tool Count4/5

    With 9 tools, the count is reasonable for a GCP server, covering key areas like billing, logging, and resource management. It's slightly lean but well-scoped, avoiding bloat while providing essential functionality.

    Completeness3/5

    The toolset covers listing and getting operations well but has notable gaps in CRUD coverage, such as missing create, update, or delete tools for resources like GKE clusters or SQL instances. This could limit agent workflows that require full lifecycle management.

    Maintenance

    ActivityInactive
    ResponsivenessOutdated data

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      B
      quality
      D
      maintenance
      Enables AI assistants to interact with and manage Google Cloud Platform resources including Compute Engine, Cloud Run, Storage, BigQuery, and other GCP services through a standardized MCP interface.
      1
      6
      MIT
    • A
      license
      C
      quality
      D
      maintenance
      Enables interaction with Google Cloud services including billing cost analysis, log querying, and metrics monitoring through natural language commands. Provides comprehensive tools for managing GCP resources, analyzing costs, detecting anomalies, and retrieving operational insights.
      40
      1
      Apache 2.0
    • F
      license
      Not graded
      quality
      D
      maintenance
      Enables managing Google Cloud Platform infrastructure through natural language, including VM deployment, SSH key management, remote command execution, and Terraform infrastructure-as-code operations.
      2
      -
    • A
      license
      B
      quality
      Not graded
      maintenance
      Enables AI assistants to interact with and manage Google Cloud Platform resources including Artifact Registry, BigQuery, Cloud Build, Compute Engine, Cloud Run, Cloud Storage, and monitoring services through a standardized MCP interface.
      1
      MIT