This server is an MCP tool that lets you query and analyze your AWS billing and cost data through natural-language-driven tool calls.
Cost & usage analysis: query spend over any date range, group by service, region, usage type, tag, etc., with filters and automatic pagination.
Dimension discovery: list actual cost-incurring values (e.g. services, regions, usage types) to build precise filters.
Cost forecasting: predict future spend with an 80% confidence interval.
Anomaly detection: retrieve AWS Cost Anomaly Detection findings and root causes.
Budget monitoring: list budgets with actual and forecasted spend.
Cost allocation tags: enumerate tag keys/values for tag-based cost grouping.
Commitment utilization: check Savings Plans and Reserved Instance utilization/coverage.
Account safety: optionally verify the AWS account identity before running queries.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@aws-billing-mcp Show me this month's cost by service, top 10."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
MCP server (stdio) — accesses Billing and Cost Management data for <目标AWS 账号> via the local AWS CLI, for MCP clients such as Quick Desktop to perform cost and usage analysis.
Node.js ≥ 18 (v24 is installed on this machine)
AWS CLI v2, and the credentials of the default profile (or the profile specified by AWS_PROFILE) belong to <目标AWS 账号>
IAM permissions: ce:Get*, budgets:ViewBudget, sts:GetCallerIdentity
On the first tool invocation, the server uses sts get-caller-identity to verify that the current credentials indeed belong to <目标AWS 账号>. If they do not match, it immediately returns an error, preventing accidental queries against other accounts.
Related MCP server: AWS FinOps MCP Server
cd ~/aws-billing-mcp
npm installnode test-client.mjsIt will list all tools, query the Top 5 costs for this month grouped by service, and run a month-end cost forecast.
Add the following to Quick Desktop's MCP server configuration (Settings → Integrations/MCP servers, or its JSON configuration file):
{
"mcpServers": {
"aws-billing": {
"command": "node",
"args": ["~/aws-billing-mcp/index.mjs"],
"env": {
"AWS_PROFILE": "default"
}
}
}
}Note: node is installed via nvm (
~/.nvm/versions/node/v24.8.0/bin/node). If Quick Desktop cannot findnodeorawsat startup, replacecommandwith the absolute path to node, and add"PATH": "/usr/local/bin:/usr/bin:/bin:/opt/homebrew/bin"toenv(include the directory containing aws).
Tool | Description |
| Core query: cost/usage for any time period, supports grouping by SERVICE, USAGE_TYPE, REGION, TAG, etc. (up to 2), supports Cost Explorer filter expressions, automatic pagination |
| Enumerates the values that actually incur costs for a dimension (service names, regions, etc.), used to determine precise filter values |
| Cost forecast for future time periods (includes 80% confidence interval) |
| Cost anomalies detected by Cost Anomaly Detection and their root causes |
| Budgets under the account and actual/forecast execution status |
| Cost allocation tag key/value in billing data, used with TAG grouping |
| Utilization and coverage of Savings Plans / RI |
“How much did this account spend this month? Rank the Top 10 by service.”
“What is the daily Bedrock consumption trend for the last 30 days? Are there any abnormal spikes?”
“Forecast the total bill at the end of August.”
“Break down EC2 costs by usage type — which part is growing the fastest?”
Variable | Default | Description |
|
| AWS account allowed to access; reject if credentials do not match |
| (system default) | AWS CLI profile to use |
|
| Path to the AWS CLI executable |
List AWS Budgets configured for the account, with limits, actual spend and forecast against each budget.
Inputs: no named parameters.
Get Savings Plans or Reserved Instance utilization and coverage for a time period. Returns an explanatory message if the account has no such commitments.
Inputs: end (required), start (required), commitment_type (required), report, +1 more.
List cost allocation tag values seen in billing data for a time period. Provide tag_key to list its values; omit it to list available tag keys. Use these keys with group_by type=TAG in get_cost_and_usage.
Inputs: end (required), start (required), tag_key.
Query AWS cost and usage for account 135709585800 over a time period. Supports grouping (e.g. by SERVICE, USAGE_TYPE, REGION, LINKED_ACCOUNT, or a cost allocation TAG) and Cost Explorer filter expressions. Start date is inclusive, end date is exclusive. This is the primary tool for spend analysis: monthly trends, per-service breakdowns, daily spikes, etc.
Inputs: end (required), start (required), filter, metrics, +2 more.
List cost anomalies detected by AWS Cost Anomaly Detection in a date range, including root causes and impact. Returns an empty list if no anomaly monitors are configured.
Inputs: end (required), start (required), max_results.
Forecast future AWS spend for the account over a date range (must start today or later). Returns the mean forecast plus an 80% prediction interval.
Inputs: end (required), start (required), filter, metric, +1 more.
List available values for a Cost Explorer dimension in a time period — e.g. which services, regions, usage types, or linked accounts actually incurred cost. Use this to discover exact value strings before filtering in get_cost_and_usage.
Inputs: end (required), start (required), dimension (required), search_string.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 7 tools
Each tool serves a distinct purpose: querying cost/usage, discovering dimension values, forecasting, detecting anomalies, listing budgets, managing cost allocation tags, and checking commitment utilization. There is no overlap or ambiguity between them.
All tool names follow a consistent verb_noun pattern in snake_case, using retrieval verbs like get_ and describe_. The pattern is uniform across the set, making it predictable for an agent.
With 7 tools, the server is well-scoped for its purpose of AWS billing and cost analysis. Each tool addresses a key aspect of cost management without being bloated or sparse.
The tool surface covers the core cost management workflows: historical analysis, forecasting, anomaly detection, budget tracking, tag discovery, and commitment utilization. There are no obvious dead ends or missing critical operations for a read-only billing server.
AWS Cost Explorer cost, forecast, and anomaly reporting through user-connected IAM credentials.
Hosted MCP server for AWS cloud spend: service breakdowns, anomalies, savings and forecasts.
FinOps assistant: query AWS, GCP and Azure cost, commitments, budgets and reports.
Remote MCP for Copilot CLI switch gate MCP, structured receipts, audit logs, and reviewer-ready evid