ImpactOps AI — AI skill for Claude Code

ImpactOps AI — Sustainable Operations Intelligence & Autonomous Decision Agent

**Hackathon Theme:** Build Beyond AI: Engineer Secure, Intelligent and Autonomous Applications **Status:** Production-Ready Architecture & Local Verifiable Prototype **Stack:** TypeScript, React 19, Vite, Express, Google Gemini 1.5 Flash, Google Stitch MCP Design Tokens

[](#automated-validation) [](LICENSE) [](#google-technologies) [](#security-engineering)

---

🌍 The Problem

Industrial manufacturing plants, hyperscale data centers, and corporate facilities waste millions of dollars and emit thousands of metric tons of avoidable greenhouse gases annually because:

1. Telemetry is fragmented across isolated building management systems (BMS), SCADA loops, and sub-meters.
2. Inefficiencies are invisible — subtle mechanical faults (like stuck bypass valves or chiller resonance locks) operate unflagged for weeks.
3. Interventions lack prioritization — maintenance teams cannot calculate financial ROI, payback periods, or carbon trade-offs on the fly.
4. Autonomous AI is treated with skepticism — facilities managers cannot trust black-box LLMs that make unverified claims or execute unauthorized hardware changes.

---

⚡ The Solution: ImpactOps AI

**ImpactOps AI** is an operational intelligence and autonomous decision agent that converts raw telemetry into explainable, verified, human-authorized sustainability interventions:

1. Continuous Anomaly Detection: Ingests multi-day streaming telemetry (electricity kWh, water kL, waste tons) and detects deviations using rolling standard deviations and statistical z-sc