GitHub Reddit (AMA!) YouTube Roadmap Download

Your home automation should work when the internet doesn't.

Krill is the first decentralized control platform where every sensor, actuator, camera, and logic rule is a node in a self-healing mesh — running on your Raspberry Pis, owned by you, with native apps on every platform.

Built from the ground up in Kotlin Multiplatform. One codebase. Android, iOS, Desktop, Web, and Raspberry Pi. No cloud. No subscription. No single point of failure.

Why Krill

🌐

Zero Cloud

Your data never leaves your network. No accounts, no subscriptions, no vendor lock-in.

🕸

Mesh Swarm

Servers auto-discover each other on your LAN. Add a Pi, enter a 4-digit PIN, and it joins the swarm.

📱

Every Platform

Native apps for Android, iOS, Mac, Windows, Linux, and web.

⚡

Real Power

Logic gates, Python lambdas, serial sensors, MQTT, GPIO, cameras, color sensors, and 35+ node types — wired together visually.

🎨

Your Dashboard

A unique force-graph shows live alerts and executions — or design custom SVG dashboards in Inkscape with real-time data and live video overlays.

Everything in Krill is a Node. Nodes react to data changes, execute logic, and control hardware — across any number of servers, in real time.

Data Sources	Triggers	Processors	Filters
Serial Devices	CRON Timers	Calculations	High Threshold
GPIO Pins	Python Lambda	Python Lambda	Low Threshold
Data Points	High / Low Alerts	Cameras	Deadbands
Web Hooks	Logic Gates	Data Compression	Debounce
MQTT	MQTT	Notifications	Data TTLs

How Nodes, Verbs, and Sources Work

The mental model behind a Krill swarm — every node is independent, observes its sources, and acts on the verb it receives using only its own capabilities. No more "executing children"; nodes are in...

May 18, 2026 Guide

How I use multiple agents for CI/CD on a complex Kotlin Multiplatform Project

How my staff of Claude agents runs a full spec-driven development pipeline for Krill, my Kotlin Multiplatform IoT control system, with zero human labor on the implementation of features and fixes.

Apr 29, 2026 Guide

Krill MCP & Claude Skill on macOS

Wire Claude Desktop on macOS up to a local krill-mcp server via the mcp-remote stdio bridge, and install the companion Krill Claude skill so every project gets it.

Apr 29, 2026 Download

Logging Vivarium Temperature and Humidity with an SHT30 and Two Lambdas

A QT Py reads an SHT30 over I2C, posts JSON to a Krill DataPoint, and two tiny Python Lambdas fan that one snapshot out into a TempF DataPoint (with C→F conversion) and a Humidity DataPoint.

Apr 28, 2026 Automate