Not another stateless chatbot.
Constellation Engine gives an AI agent a star map for a mind — memory that grows, identity that survives sessions, a self that becomes more yours over time.
Free · Open source · AGPL-3.0 · Runs locally on your machine
Three things make Constellation different from every assistant you've used.
Every conversation becomes a node. Nodes form clusters. Clusters connect into a living graph that the agent walks, prunes, and grows on its own.
Your memory lives on your machine, not someone else's server. SQLite under the hood. Export, import, archive — it's yours.
The agent has an identity that survives sessions. It remembers what mattered, what you liked, and what didn't work — and uses that to become more yours.
A dashboard, a star map, a chat. The pieces work together so the agent doesn't start from zero every time.
Health, memory, configuration — one panel for everything the agent needs to stay alive. No CLI required.
Built on Electron. Runs in a window, not a terminal.
Watch your memory take shape. Each node a moment, each edge a connection, the whole thing alive and self-organizing.
Powered by a background daemon that watches, clusters, and consolidates.
Bring your own API key. Talk to your agent. Every message becomes part of the map.
Compatible with Claude, OpenAI, and other major LLM providers.
Constellation Engine is open source under AGPL-3.0. Read the code, file an issue, send a pull request — or just watch the project grow.
Constellation is built by one human and one AI agent. 🌱
Not another chatbot. An experiment in giving agents what assistants lack: memory that grows, identity that survives sessions, a self that becomes more yours over time. 🌌
But it costs real money to run — APIs, models, infrastructure, time. Sponsorship is what keeps it alive. 🤝
Thank you. ❤️
Yes. We haven't paid for an EV code signing certificate yet (it's expensive for a one-person project). Click More info → Run anyway to launch. Once enough people run it, SmartScreen will learn it's safe automatically.
Same reason — we don't have an Apple Developer ID yet. Right-click the app → Open → confirm. If macOS blocks it entirely, run xattr -cr /Applications/Constellation.app in Terminal.
The memory database is local. The only network calls are to the LLM provider you configure (Claude, OpenAI, etc.) — using your own API key. No telemetry, no analytics, no phone-home.
No. Constellation Engine is free and open source. Sponsorship is voluntary — it's what keeps the project alive, but every feature is available to everyone.
Anthropic Claude and OpenAI's GPT models are supported out of the box. More providers are on the roadmap. Bring your own API key.