Library / Monitoring, Ops & Incidents
Monitoring, Ops & Incidents
Detect AWS and GCP cost anomalies with GPT-4o-mini and alert Slack
Runs every morning, compares yesterday's AWS and GCP spend with recent history, flags spikes past dollar, percent and z-score thresholds, and sends AI-summarized alerts to Slack.
What it does
## 💸 FinOps – Cloud Cost Anomaly Agent (AWS + GCP) For engineering and finance teams who find out about a runaway NAT gateway, a forgotten GPU instance or a BigQuery job gone wild when the monthly bill lands, weeks after the money is spent. ### How it works Every morning the workflow pulls yesterday's spend by service and region from AWS Cost Explorer and your GCP billing export. It compares each line with the previous day and the past week, and flags only real spikes that pass a dollar floor, a percentage jump and a z-score. For each AWS spike it pulls recent CloudTrail changes in that region. GPT-4o-mini combines them with the usage-type breakdown to suggest the most likely cause. Slack gets an alert with the cost, monthly impact, probable cause and two buttons: mark it expected, or ignore it for seven days. Every anomaly and decision is logged in Google Sheets. ### Setup steps 1. Create a Google Sheet with an Anomaly_Log tab. 2. Add AWS credentials with ce:GetCostAndUsage and cloudtrail:LookupEvents, and Google OAuth2 with BigQuery read for GCP. 3. Connect Slack and OpenAI. 4. In Set Cost Agent Config, turn each cloud on or off and set the GCP project, billing table, thresholds and your n8n base URL (used for the buttons). 5. Run once manually, then activate.
The steps
- Every Day at 9 AM IST
- Set Cost Agent Config
- Read Anomaly Log Sheet
- AWS Enabled?
- Get AWS Cost by Service & Region
- Get AWS Cost by Usage Type
- Normalize AWS Costs
- GCP Enabled?
- Query GCP Billing Export
- Normalize GCP Costs
- Detect Day-over-Day Anomalies
- Loop Over Anomalies
- Build Run Summary
- Post Run Summary to Slack
29 steps in total.
Similar automations
MCP Servers & ToolsAWS Cost and Usage Reports as an MCP Server5 stepsMonitoring, Ops & IncidentsDetect ad spend anomalies and pacing issues across Google Ads, Meta and GA423 stepsImage, Video & AudioTranscribe Audio from AWS S3 with AWS Transcribe4 stepsAlerts & NotificationsDetect and investigate subscription revenue anomalies with Postgres and Claude34 steps