finops-analysis

name	finops-analysis
description	AWS FinOps analysis — cost breakdown, trends, forecasts, anomalies, and optimization recommendations. Three paths: (1) MCP tools via deployed gateway (preferred); (2) direct AWS CLI; (3) delegation to coding agent. Produces markdown tables and summaries. Use when the user asks about AWS costs, spend, budget, savings, optimization, or cost trends.
argument-hint	[what do you want to know? e.g. 'top services by cost', 'cost forecast', 'savings recommendations']
allowed-tools	Bash, Write, Read, Glob, Grep
user-invocable	true

AWS FinOps Analysis

Analyze AWS costs, produce spend breakdowns, identify trends, surface anomalies, and recommend optimizations. Output is markdown — tables, summaries, and actionable recommendations.

Routing

Are cost-explorer / billing / cost-optimization-hub MCP tools available?
├── Yes → Path M (use MCP tools — fastest, handles cross-account)
└── No
    ├── Can you run `aws --version`?
    │   ├── Yes → Path A (AWS CLI directly)
    │   └── No
    │       ├── Is a coding agent enabled? → Path B (delegate)
    │       └── No → Stop. Tell the user to deploy with DEPLOY_MODE=gateway-only
    │                    or run in Claude Code / Kiro.

Path M — MCP Tools (preferred)

When these tools are available, use them directly:

Tool	Use for
get_cost_and_usage	Spend by service, account, region, tag. Monthly or daily granularity
get_cost_and_usage_comparisons	Month-over-month or period-over-period changes
get_cost_forecast	Projected spend for current/next month
get_dimension_values	List available services, accounts, regions for filtering
get_anomalies	Cost anomalies detected by AWS Cost Anomaly Detection
get_billing_alerts	Active budgets and their current vs limit status
get_account_info	Account ID, org membership, billing details
get_enrollment_status	Check if Cost Optimization Hub is enabled
list_recommendations	Right-sizing, idle resources, savings plans, reserved instances
list_recommendation_summaries	Aggregated savings by resource type or region

Workflow

1. Understand the question — classify: spend breakdown, trend, forecast, anomaly check, or optimization
2. Call the appropriate tools — don't over-fetch. One or two tool calls usually suffice
3. Format as markdown — tables for data, bullet points for insights, bold for key numbers

Example flows

"What are my top services by cost this month?" → get_cost_and_usage with Granularity=MONTHLY, GroupBy=SERVICE, current month

"How does this month compare to last?" → get_cost_and_usage_comparisons comparing current to prior month

"Any savings opportunities?" → get_enrollment_status first (check COH is enabled), then list_recommendations

"Cost forecast?" → get_cost_forecast for current month

Path A — AWS CLI

If MCP tools aren't available but AWS CLI is:

# Top services by cost (current month)
aws ce get-cost-and-usage \
  --time-period Start=$(date -u +%Y-%m-01),End=$(date -u +%Y-%m-%d) \
  --granularity MONTHLY \
  --metrics AmortizedCost \
  --group-by Type=DIMENSION,Key=SERVICE

# Forecast
aws ce get-cost-forecast \
  --time-period Start=$(date -u +%Y-%m-%d),End=$(date -u +%Y-%m-01 -d "+1 month") \
  --granularity MONTHLY \
  --metric AMORTIZED_COST

# Anomalies (last 30 days)
aws ce get-anomalies \
  --date-interval Start=$(date -u -d "-30 days" +%Y-%m-%d),End=$(date -u +%Y-%m-%d)

# Optimization recommendations
aws cost-optimization-hub list-recommendations --max-results 20

Parse JSON outputs and format as markdown tables.

Path B — Delegate

Hand the coding agent this brief:

"Run AWS Cost Explorer CLI commands to answer: [user's question]. Return the results as markdown tables with cost figures in USD. Include month-over-month comparison if relevant."

Output format

Always produce:

• Summary line — one sentence with the key number (total spend, top service, savings available)
• Data table — ranked, with columns for service/account, cost, change
• Insights — 2-3 bullet points on what stands out (spikes, anomalies, trends)
• Recommendations — if optimization data is available, prioritized list

Output Template

# FinOps Analysis

**Period:** [month/date range]
**Account:** [account ID or "Organization"]
**Generated:** [timestamp]

## Spend Summary

| Metric | Value |
|--------|-------|
| Total spend (period) | $[X,XXX.XX] |
| Month-over-month change | [+/-X.X]% |
| Forecast (end of month) | $[X,XXX.XX] |
| Top service | [service] ($[X,XXX]) |

## Top Services by Cost

| # | Service | Cost (USD) | MoM Change | % of Total |
|---|---------|-----------|-----------|-----------|
| 1 | [service] | $[X,XXX] | [+/-X]% | [X]% |
| 2 | [service] | $[X,XXX] | [+/-X]% | [X]% |

## Anomalies

| Service | Detected | Expected | Actual | Impact |
|---------|----------|----------|--------|--------|
| [service] | [date] | $[X] | $[X] | +$[X] |

_(or "No anomalies detected in this period")_

## Optimization Opportunities

| # | Resource | Type | Action | Monthly Savings |
|---|----------|------|--------|----------------|
| 1 | [id] | [type] | [right-size/delete/reserve] | $[X] |

**Total identified savings:** $[X,XXX]/month

## Key Insights

- [2-3 bullet points on what stands out]

Adapt the template to the user's question — don't generate all sections if they only asked about top services.

Constraints

• Never fabricate cost figures. All numbers must come from tool/CLI output
• Use AmortizedCost by default (includes RI/SP amortization) unless user asks for UnblendedCost
• Dates: use ISO format (YYYY-MM-DD). Current month = 1st of month to today
• If a tool returns empty data, say so honestly — don't invent numbers