Open-source cloud cost detective: pandas does the math, Gemini explains

ZtasnimS75 · reddit · 2026-10-31

A developer released AI Cloud Cost Detective, an open-source tool that analyzes cloud billing CSVs for cost spikes: daily trends, service/region breakdowns, anomalies grouped into incidents, savings hints, and a 30-day linear forecast.

Key design: all numbers come from pandas, never an LLM; Gemini only writes a plain-English explanation per incident, with a template fallback if it errors or rate-limits.

Anomaly detection: rolling 14-day baseline with z-score >= 3 and >= 1.5x, plus a "new service" rule. Flagged days are excluded from the baseline, fixing a bug where multi-day spikes hid themselves after day two.

Testing: on FinOps Foundation FOCUS sample data (100k rows, AWS+Azure), it flagged a 9-day EC2 incident $428 above baseline. 25/25 regression tests pass. Known limits: sample data only, fixed thresholds can false-alarm on cheap services, credits dropped, no precision/recall measured. Demo and code are public.

Original post →

More from coding & agent

coding & agent channel →