AWS FinOps MCP Server
Enables natural language analysis of AWS costs, automated FinOps waste audits, and budget monitoring across multiple profiles and regions while keeping credentials secure locally.
README
AWS FinOps MCP Server
An MCP (Model Context Protocol) server that brings powerful AWS FinOps capabilities directly into your AI assistant. Analyze cloud costs, audit for waste, and get budget insights using natural language, all while keeping your credentials secure on your local machine.
Why Use the AWS FinOps MCP Server?
Managing AWS costs can be complex. Finding answers to questions like "What did we spend on S3 last month?" or "Do we have any unused EC2 instances?" often requires navigating the AWS Console or writing scripts.
This server bridges that gap by connecting the powerful, natural language understanding of an LLM directly to your AWS account's financial data. You can ask complex questions and get immediate, actionable insights without ever leaving your chat interface.
Use Cases:
Prompt on Claude Desktop: "Get
usage-typebased cost for all my AWS CLI profiles for the last 90 days. And get theaudit reportfor all profiles for regionap-south-1. Make a finops report with data in minimalistic way and keep it professional."

Prompt on Amazon Q CLI: "Could you use
aws-finopsmcp server and get cost data for each month in the last 6 months forprofile 04and present the data in a tabular format highlighting how the total cost and cost for each service has changed every month?"

Prompt on Claude Desktop: "Generate a comprehensive finops report for management using all the tools available in the AWS FinOps MCP server. Use all the available CLI profiles and use the region ap-south-1 wherever required."

Other sample reports generated by Claude using AWS FinOps MCP Server
<a href="https://ravikiranvm.github.io/finops-usage-report/" target="_blank" rel="noopener noreferrer">Network Usage & Cost Analysis Report</a>
<a href="https://ravikiranvm.github.io/finops-comprehensive/" target="_blank" rel="noopener noreferrer">Comprehensive FinOps Report</a>
Table of Contents
- What is an MCP Server?
- Why Use the AWS FinOps MCP Server?
- Key Features
- Prerequisites
- Installation
- Configuration
- Usage & Example Prompts
- Tool Reference
- Cost Considerations
- Development & Contributing
- License
Key Features
- Detailed Cost Analysis: Query your AWS cost data using tags, time ranges, and service groupings.
- Automated FinOps Audit: Instantly find common sources of cloud waste, such as stopped EC2 instances, unattached EBS volumes, and unassociated Elastic IPs etc.
- Budget Monitoring: Check the status of your AWS Budgets to see if you are on track, over budget, or forecasted to exceed your limits.
- Multi-Profile & Multi-Region: Seamlessly query across any or all of your configured AWS accounts and regions in a single command.
- Secure by Design: Your AWS credentials never leave your machine. The server runs locally and uses your existing AWS CLI configuration to make API calls directly to AWS.
Prerequisites
General Requirements
- Python 3.10+: Ensure you have a compatible Python version installed.
- MCP Client: An application that supports MCP. This guide uses Claude for Desktop as the example.
AWS Requirements & IAM Permissions
- AWS CLI: The AWS CLI must be installed on your machine.
- AWS Account & Profile(s): You need at least one configured AWS profile.
- IAM Permissions: The AWS identity (IAM User or Role) associated with your profile(s) requires specific read-only permissions to function correctly.
Create an IAM policy with the following JSON and attach it to your user or role. These permissions are read-only and grant access only to the necessary services.
{
"Version": "2012-10-17",
"Statement": [
{
"Sid": "AllowFinOpsMCPServer",
"Effect": "Allow",
"Action": [
"ce:GetCostAndUsage",
"budgets:ViewBudget",
"ec2:DescribeInstances",
"ec2:DescribeVolumes",
"ec2:DescribeAddresses",
"sts:GetCallerIdentity"
],
"Resource": "*"
}
]
}
Installation
There are several ways to install the AWS FinOps MCP Server:
Option 1: Using pipx (Recommended)
pipx install aws-finops-mcp-server
If you don't have pipx, install it with:
python -m pip install --user pipx
python -m pipx ensurepath
Option 2: Using uv (Fast Python Package Installer)
uv is a modern Python package installer and resolver that's extremely fast.
# Install uv if you don't have it yet
curl -LsSf https://astral.sh/uv/install.sh | sh
# Create project directory
mkdir aws-finops-mcp-server && cd aws-finops-mcp-server
# Create and activate python virtual environment
uv venv && source .venv/bin/activate
# Install aws-finops-dashboard
uv pip install aws-finops-mcp-server
Option 3: From Source
# Clone the repository
git clone https://github.com/ravikiranvm/aws-finops-mcp-server.git
cd aws-finops-mcp-server
# Create and activate python virtual environment
python -m venv .venv && source .venv/bin/activate
# Install using pip
pip install -e .
Configuration
Step 1: AWS CLI Profile Setup
If you haven't already, configure your AWS profiles. For each profile you want to use, run:
aws configure --profile your-profile-name
The CLI will prompt you for:
AWS Access Key ID: Your credential's access key.AWS Secret Access Key: Your credential's secret key.Default region name: e.g.,us-east-1.Default output format: e.g.,json.
For your default profile, you can omit the --profile flag.
Step 2: MCP Client Setup (Claude for Desktop)
To use this server with Claude for Desktop, you need to tell the application how to run it.
-
Open your Claude for Desktop configuration file. It's located at:
- macOS / Linux:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
- macOS / Linux:
-
Add the following entry to the
mcpServersobject. If the file doesn't exist or is empty, you'll need to create the JSON structure.{ "mcpServers": { "aws_finops": { "command": "aws-finops-mcp-server", "args": [] } } } -
Save the file and completely restart Claude for Desktop.
After restarting, the server will launch in the background, and its tools will be available to Claude.
⚠️ Note:
After installing this tool (e.g., viapipx install aws-finops-mcp-server), ensure theaws-finops-mcp-servercommand is available in your system'sPATH.To check where it's installed, run:
which aws-finops-mcp-serverIf it returns a path like
/Users/yourname/.local/bin/aws-finops-mcp-server, but it's not in your PATH, you have two options:
- Run
pipx ensurepathand restart your terminal, or- Use the full path in your
mcpServersconfig:{ "mcpServers": { "aws_finops": { "command": "/full/path/to/aws-finops-mcp-server", "args": [] } } }
Usage & Example Prompts
You can now ask questions about your AWS finances in plain English.
Use Case: Cost & Usage Analysis
Basic Query: "Show me my AWS costs for the
defaultprofile this month."
Time-Range Query: "get aws cost for my
devprofile for the last 14 days"
Filtered & Grouped Query: "What was the cost for my
productionprofile last month, filtered by the tagCostCenter=Project-Alpha, and grouped byINSTANCE_TYPE?"
Multi-Profile Query: "Compare the costs for my
stagingandproductionprofiles over the last 30 days."
Use Case: FinOps Waste & Savings Audit
Simple Audit: "Run a finops audit on my
defaultprofile inus-east-1."
Multi-Region, Multi-Profile Audit: "Check for unused resources in
us-west-2andeu-west-1for myprod-usandprod-euprofiles."
Comprehensive Audit for All Profiles: "Run a FinOps audit for all my configured profiles in
us-east-1."
Supported Clients
- Claude Desktop
- Amazon Q CLI
- Any MCP-compatible client supporting tools
View the list of clients that support MCP Servers at https://modelcontextprotocol.io/clients
Tool Reference
get_cost
Fetches cost and usage data from AWS Cost Explorer.
| Parameter | Type | Description |
|---|---|---|
profiles |
List[str] |
A list of AWS CLI profile names to use. |
all_profiles |
bool |
If True, use all available profiles. Defaults to False. |
time_range_days |
int |
Number of past days to fetch data for (e.g., 7 for last 7 days). |
start_date_iso |
str |
Start date in YYYY-MM-DD format. Takes precedence over time_range_days. |
end_date_iso |
str |
End date in YYYY-MM-DD format. Must be used with start_date_iso. |
tags |
List[str] |
Filter by cost allocation tags. Format: ["Key=Value", "Key2=Value2"]. |
dimensions |
List[str] |
Filter by dimensions. Format: ["REGION=us-east-1"]. |
group_by |
str |
The dimension to group costs by. Default is SERVICE. |
run_finops_audit
Runs a financial audit to find unused and potentially costly resources.
| Parameter | Type | Description |
|---|---|---|
regions |
List[str] |
A list of AWS regions to run the audit in (e.g., ["us-east-1", "eu-west-1"]). |
profiles |
List[str] |
A list of AWS CLI profile names to use. |
all_profiles |
bool |
If True, use all available profiles. Defaults to False. |
Output Format: JSON, CSV, PDF, HTML, Markdown (depending on client capabilities)
Cost Considerations
This tool interacts with the AWS Cost Explorer API. AWS charges for these API calls.
- Cost: Every time you use
get_costtool, it makes 2 calls to the Cost Explorer API and it costs $0.01. - The
run_finops_audittool calls other AWS service APIs (EC2, Budgets) which are generally included in the free tier or have negligible costs at typical usage levels.
Development & Contributing
Contributions are welcome! If you'd like to improve the server or add new features, please follow these steps:
- Fork and Clone: Fork the repository and clone it to your local machine.
- Set Up Environment:
git clone https://github.com/your-username/aws-finops-mcp-server.git cd aws-finops-mcp-server python -m venv .venv source .venv/bin/activate # On Windows, use `.venv\Scripts\activate` - Install in Editable Mode:
pip install -e . - Make Changes: Add your features or bug fixes.
- Submit a Pull Request: Open a PR against the
mainbranch with a clear description of your changes.
License
This project is licensed under the MIT License. See the LICENSE file for full details.
This is a sister project of https://github.com/ravikiranvm/aws-finops-dashboard
Made with ❤️ by Ravi Kiran
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。