Cloud FinOps Skill & MCP — Claude Code MCP server
Supplies FinOps playbooks and cloud billing knowledge across AWS, Azure, GCP, and OCI to AI agents. Helps audit infrastructure costs and estimate AI model inference spend.
claude mcp add --transport http cloud-finops https://mcp.optimnow.io/mcpWhat Cloud FinOps Skill & MCP does
This MCP server exposes six retrieval tools focused on cloud spend management across AWS, Azure, GCP, and OCI. It provides structured guidance on resource allocation, Kubernetes workloads, database commitments, and chargeback strategies. Instead of guessing billing nuances, the assistant pulls verified FinOps methodologies directly into the chat session.
Standard AI models frequently miscalculate inference economics and confuse provider-specific discount terms. The server prevents those errors by supplying tested waste playbooks and calculation formulas on demand. No vector database setup is needed.
Deployment options include a hosted HTTP endpoint and a local Python package via cloud-finops-mcp. For the hosted service, the URL https://mcp.optimnow.io/mcp must not have a trailing slash, as the client sandbox requires the exact path.
What it helps with
- Calculate break-even thresholds for Provisioned Throughput Units (PTU)
- Compare capacity reservation mechanisms across AWS and Azure
- Detect hidden network spend caused by NAT gateway traffic
- Structure cloud chargeback and cost allocation frameworks
Example prompts for the agent
- How can we route traffic from NAT Gateway to S3 to reduce our data transfer bill?
- Calculate the break-even threshold for PTU given our current inference load.
- Explain the differences between AWS Savings Plans and Azure reservations for our database.
How to install Cloud FinOps Skill & MCP
1Platform CLI: remote server
claude mcp add --transport http cloud-finops https://mcp.optimnow.io/mcp2Platform CLI
claude mcp add cloud-finops-skill-mcp -- uvx cloud-finops-mcpuvx runs the command named after the package; if the README names another one, use uvx --from cloud-finops-mcp <command>
1Platform CLI: remote server
codex mcp add cloud-finops-skill-mcp --url https://mcp.optimnow.io/mcp2Platform CLI
codex mcp add cloud-finops-skill-mcp -- uvx cloud-finops-mcpuvx runs the command named after the package; if the README names another one, use uvx --from cloud-finops-mcp <command>
Config file
{
"mcpServers": {
"cloud-finops-skill-mcp": {
"url": "https://mcp.optimnow.io/mcp"
}
}
}{
"mcpServers": {
"cloud-finops-skill-mcp": {
"command": "uvx",
"args": [
"cloud-finops-mcp"
]
}
}
}uvx runs the command named after the package; if the README names another one, use uvx --from cloud-finops-mcp <command>
Config file
{
"servers": {
"cloud-finops-skill-mcp": {
"type": "http",
"url": "https://mcp.optimnow.io/mcp"
}
}
}{
"servers": {
"cloud-finops-skill-mcp": {
"type": "stdio",
"command": "uvx",
"args": [
"cloud-finops-mcp"
]
}
}
}uvx runs the command named after the package; if the README names another one, use uvx --from cloud-finops-mcp <command>
1Platform CLI: remote server
gemini mcp add -t http cloud-finops-skill-mcp https://mcp.optimnow.io/mcp2Platform CLI
gemini mcp add cloud-finops-skill-mcp uvx cloud-finops-mcpuvx runs the command named after the package; if the README names another one, use uvx --from cloud-finops-mcp <command>
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Cloud FinOps Skill & MCP
How do I install Cloud FinOps Skill & MCP in Claude Code?
Run claude mcp add --transport http cloud-finops https://mcp.optimnow.io/mcp to connect to the hosted remote server. For a local setup, run claude mcp add cloud-finops-skill-mcp -- uvx cloud-finops-mcp instead.
What does Cloud FinOps Skill & MCP do?
It gives AI agents six search tools covering cloud cost optimization across AWS, Azure, GCP, and OCI. The server answers technical queries regarding waste detection, chargeback structures, and AI model inference economics.
Do I need to install a local Python package to use this server?
A local Python installation is not required because OptimNow provides a hosted HTTP endpoint. When configuring the connection, pass https://mcp.optimnow.io/mcp exactly without a trailing slash.
