CloudScope MCP is a multi-cloud cost management MCP server that supports Azure and GCP, designed to let users directly query and analyze cloud bills through an AI assistant. It provides read-only access, covering spend, anomalies, forecasts, budgets, and guided FinOps workflows.
Purpose
CloudScope's core purpose is to simplify cloud cost management. Users only need to ask the AI a question, such as "give me this month's cost review," and CloudScope will run the complete workflow (spend, changes, anomalies, idle resources, budgets, forecasts) and return a decision-ready report, rather than raw numbers that users must interpret themselves. The sample report shows monthly Azure spend, major changes, idle resource costs, budget usage, and recommended actions.
Key Capabilities
- Spend analysis: view cloud spend for the current month or a specified time period.
- Anomaly detection: identify cost anomalies.
- Forecasting: predict future costs.
- Budget tracking: monitor budget usage progress.
- Idle resource detection: discover unused or wasted resources.
- Guided FinOps workflows: generate structured reports through predefined prompts.
- Cross-account queries: support cost comparisons across subscriptions (Azure) and across projects (GCP).
How to Use
Prerequisites
- Azure: install and sign in to the Azure CLI (az login), and have the Cost Management Reader role on the subscription.
- GCP: install and sign in to the gcloud CLI (gcloud auth application-default login), enable BigQuery billing export (one-time manual setup), and have the BigQuery Data Viewer and BigQuery Job User roles on the dataset project.
Configuration
- Azure zero configuration: CloudScope automatically detects the subscription from az login. Just add the server. To specify a subscription, add the environment variable AZURE_SUBSCRIPTION_ID.
- GCP: requires a BigQuery billing export table. Find the table name in the GCP Console > Billing > Billing Export > BigQuery Export. Set GOOGLE_CLOUD_PROJECT to the GCP project ID; if the billing dataset is in a different project, you can use GCP_PROJECT_ID override.
Example questions
- "How much did Azure spend last month?"
- "Show GCP spending by service over the past 7 days"
- "Were there any cost anomalies in GCP this week?"
- "How much is Azure expected to spend next month?"
- "Show cost optimization recommendations for GCP"
- "Compare Azure and GCP costs across all projects and subscriptions"
Tools and prompts
Most tools accept a provider parameter (azure or gcp); provider-specific tools (list_subscriptions, list_projects, and the cross-account tools get_cross_subscription_costs / get_cross_project_costs) are fixed to their respective clouds. The default provider is auto-detected based on the configured provider. Guided workflows generate structured reports, and all prompts accept an optional provider parameter. In Claude Code, type /cloudscope: to view all prompts; in Claude Desktop, click the + button → Connectors → cloudscope.
When to use
- FinOps teams or individuals who need a quick overview of multi-cloud spending.
- Those who want to query cloud costs in natural language without manually operating the console.
- Scenarios that require regular cost reviews and optimization recommendations.
- Users who use both Azure and GCP and want a unified view and comparison of costs.
Limitations
- CloudScope is read-only and cannot create, modify, or delete any cloud resources.
- Azure uses the Cost Management Reader permission; GCP uses BigQuery Data Viewer + Job User, with no write permissions.
- No service principal is required; for local use, az login or gcloud auth application-default login is sufficient.
- The Azure Cost Management API is free.
- GCP BigQuery cost queries incur charges, typically less than $0.01 per query ($6.25/TiB scanned, with the first 1 TiB per month free).
- Azure can be used at the same time as GCP; just configure two sets of environment variables.
- Security policy: read-only access.
