Open-source Model Context Protocol (MCP) resources for Cloud FinOps: cloud cost optimization, AI cost management, FinOps automation, tagging governance, cost anomaly detection, and budget monitoring — across AWS, Azure, and GCP.
A practical resource hub for FinOps practitioners, cloud engineers, and platform teams who want to use AI agents to automate cloud cost optimization. This repo provides tutorials, MCP server documentation, client guides, and security frameworks for implementing the Model Context Protocol across AWS, Azure, and GCP.
18 MCP servers documented | 7 step-by-step tutorials | 9 MCP clients compared | 3 cloud providers covered
- Choose an MCP client — Claude, ChatGPT, Gemini, Copilot, Cursor, Kiro, VS Code
- Run your first MCP tutorial — query real-time AWS pricing in 15 minutes
- Explore all FinOps MCP servers — AWS, Azure, GCP, Tagging, JIRA, Slack
- New: Workflow & Collaboration Servers — FinOps Tagging Compliance, JIRA, and Slack MCP servers for automated cost governance
- Registry Backfill —
registry.yamlnow lists all 18 documented MCP servers - MCP Authentication Vulnerabilities (January 2026) — Critical security risks and remediation (read more)
- AWS MCP Remote Server — GA with 15,000+ AWS APIs (tutorial)
| Section | What you'll find |
|---|---|
| /foundations | What is MCP, how it works, architecture deep-dive |
| /servers | 18 MCP servers — AWS, Azure, GCP, Tagging, JIRA, Slack |
| /clients | 9 MCP clients compared — Claude, ChatGPT, Gemini, Copilot, Cursor, Kiro |
| /tutorials | 7 step-by-step guides for AWS, Azure, GCP cost analysis |
| /governance | Security best practices, IAM policies, vulnerability guides |
MCP connects AI clients to cloud cost management tools through a standardized protocol
The Model Context Protocol (MCP) is an open standard that lets AI agents securely connect to external tools — like AWS Cost Explorer, GCP BigQuery billing exports, Azure Cost Management, or third-party FinOps platforms like Vantage. It's the bridge between "ask a question about cloud spend" and "get a real answer from live data."
Industry Adoption (March 2026): MCP was donated to the Linux Foundation (December 2025) with backing from Anthropic, OpenAI, Google, Microsoft, and AWS. The ecosystem has grown to 10,000+ active MCP servers and is supported by all major AI platforms.
- Cost anomaly detection — AI agents query Cost Explorer and alert on spend spikes
- Tagging compliance — automated audits of tag coverage and drift detection
- Resource optimization — rightsizing recommendations, idle resource identification
- Budget monitoring — real-time threshold alerts via Slack, JIRA ticket creation
- Multi-cloud cost analysis — unified queries across AWS, Azure, and GCP
- Cost simulation — what-if scenarios for commitment purchases and architecture changes
But MCP also raises governance and security challenges — this repo addresses both.
We welcome contributions — new MCP servers, tutorials, use cases, and security guidance.
- Fork the repo and create a branch
- Add your content
- Open a PR with a clear description
See CONTRIBUTING.md for details. New to MCP or FinOps? Start with issues labeled good first issue.