cloud-cost-mcp is a Model Context Protocol (MCP) server for cloud infrastructure pricing, used for multi-cloud pricing comparison across AWS, Azure, GCP, and OCI. Users can compare compute, storage, egress, and Kubernetes costs with a single query.
Purpose The tool is mainly used for cloud cost comparison and estimation, helping users evaluate instance, storage, traffic, and Kubernetes-related costs across multiple major cloud providers, and supporting savings estimates during migration planning.
Key capabilities
- Coverage: supports 4,900+ instance types. AWS includes 1,300+ EC2 instances, 1,180+ managed database types, and Lightsail; Azure includes 1,680+ VM types (from the official Azure Retail Prices API); GCP includes 360+ instance types, covering 40+ regions; OCI includes 600+ products (via the Oracle API).
- No API key required: all data comes from public APIs (instances.vantage.sh and vendor APIs).
- Data updates: bundled data for AWS/Azure/GCP is automatically refreshed weekly via GitHub Actions, and refresh_* tools can fetch real-time pricing on demand (in-memory cache for 60 minutes). OCI bundled data is manually curated.
- Natural language queries: you can ask Claude questions such as "What is the cheapest option with 4 vCPU and 16GB?"
- Workload calculator: estimate the full workload cost including compute, storage, and egress.
- Migration planning: calculate potential savings when switching cloud providers.
Usage This server supports Claude Code and Claude Desktop users. For Claude Desktop, add the configuration to claude_desktop_config.json (macOS path: ~/Library/Application Support/Claude/claude_desktop_config.json; Windows path: %APPDATA%\Claude\claude_desktop_config.json), then restart Claude Desktop. Manual installation is also supported, and the installed binary can be referenced in any MCP client configuration. After installation, you can ask Claude directly, for example to compare compute, compare storage, compare egress, estimate full workload costs, make quick estimates, get GPU pricing (OCI), and plan migrations.
Available tools
- Comparison tools
- Calculator tools: available presets include small-web-app, medium-api-server, large-database, ml-training, kubernetes-cluster, data-lake, high-egress-cdn, high-traffic-web, gpu-inference, gpu-training-small, gpu-training-large
- Data management tools
- Real-time API tools
- GPU tools (OCI): available GPU models include A10, A100 80GB, H100 80GB, H200 141GB, L40S, MI300X All vendors now support real-time pricing refresh via public APIs. You can use check_api_status to verify API accessibility and get_data_freshness to check the freshness of bundled data.
OCI cost advantages
- 10TB free egress per month (AWS/Azure/GCP provide 100GB)
- Free Kubernetes control plane (basic clusters)
- Globally uniform pricing (no regional differences)
- Always Free tier includes 4 OCPU + 24GB RAM on ARM
- Free Kubernetes control plane: OCI (OKE) free basic clusters; Azure (AKS) free control plane; AWS (EKS) and GCP (GKE) cost $73 per cluster per month
Data model This server uses a two-layer data model:
- Bundled data (src/data/bundled/*.json) is the data read by default by the comparison/calculator tools, so the server can work instantly without network calls. AWS, Azure, and GCP bundled data is regenerated from live upstream sources (instances.vantage.sh + official Azure Retail Prices API), and via refresh-data GitHub Action automatically refreshes weekly. You can also use npm run fetch-data to regenerate locally. The OCI bundled data is manually curated (vantage.sh has no OCI data source) and manually updated. For always-real-time OCI pricing, you can use the refresh_oci_pricing tool or the dedicated oci-pricing-mcp.
- Real-time refresh tools (refresh_aws_ec2_pricing, refresh_azure_full_pricing, refresh_gcp_pricing, refresh_oci_pricing, refresh_aws_rds_pricing) fetch current pricing from vendor APIs when called and cache it in memory for 60 minutes during the process lifetime. These tools return the latest data to the caller but do not rewrite the bundled files—only npm run fetch-data / the weekly Action rewrites them. You can use get_data_freshness to view each vendor's bundle timestamp and staleness.
Use cases Suitable for users who need cloud cost comparison, workload cost estimation, migration savings assessment, and GPU pricing queries across AWS, Azure, GCP, and OCI, especially developers and cloud architects using Claude Code or Claude Desktop.
Limitations All pricing data comes from public APIs and data sources and is for informational and estimation purposes only. Cloud pricing changes frequently and may vary by region, account type, commitment terms, and other factors. Users must verify all pricing information directly with the cloud vendors before making purchasing or architectural decisions. The author makes no guarantees regarding the accuracy, completeness, or timeliness of the data.
