Analysis updated 2026-05-18
Build a personal AI workspace where chat history, saved files, and project context persist locally instead of living only in a remote server.
Create configurable AI personas with their own memory and tool permissions for different kinds of work.
Self-host your own backend so conversations and data never leave infrastructure you control.
Track every AI tool call as a visible, replayable event instead of a hidden side effect.
| aevella/polaris-local-first | md0070/polymarket-trading-bot | nextweb4/yaml-proxy-editor | |
|---|---|---|---|
| Stars | 164 | 163 | 165 |
| Language | TypeScript | TypeScript | TypeScript |
| Setup difficulty | moderate | hard | hard |
| Complexity | 4/5 | 4/5 | 3/5 |
| Audience | developer | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Frontend runs standalone with npm run dev, but connecting a real backend requires setting an API origin environment variable.
Polaris is a local-first AI workspace built for people who work with AI models over long periods of time, not just single conversations. Instead of treating a chat as one disposable message stream, Polaris treats it as a persistent workspace where conversations survive restarts, collaborators keep a consistent identity and memory, useful results turn into saved cards or project materials, and every tool the AI uses leaves a visible record that later requests can reference. The project is organized around a few core ideas. There is a chat workspace with long-lived conversations, streaming replies, attachments, and import and export flows. There are configurable collaborators, essentially AI personas with their own settings, memory controls, and permissions for which tools they can use. There is a collection area for saved conversations, code snippets, images, files, and project materials that live outside the normal chat timeline. And there is a local-first data layer called LocalData that stores all of this on the user's own device rather than a remote server by default. Polaris runs across multiple platforms from one shared codebase: web, iOS, Android, and desktop, with native bridges only where a platform genuinely needs them. On native mobile it stores data in SQLite, while web and self-hosted browser builds use browser storage like IndexedDB. The project includes a Cloudflare Worker backend package for anyone who wants to connect their own backend or self-host. Getting started locally means installing dependencies and running the Vite development server, which starts the frontend only. Connecting a real backend requires setting an environment variable to point at a self-hosted or relay API origin, with more detailed setup notes provided in the project's documentation folder. The source code is licensed under AGPL-3.0-only, and the project welcomes contributions such as bug fixes, documentation improvements, and platform bridge work, while keeping overall product direction owner-led.
Polaris is a local-first AI workspace where conversations, AI personas, tools, and saved materials persist on your own device across web, mobile, and desktop.
Mainly TypeScript. The stack also includes TypeScript, Vite, SQLite.
AGPL-3.0 requires that if you modify and run this software as a network service, you must also release your modified source code.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.