Local-first AI tools, running in your browser.
We don’t run a backend that stores your chats or credentials. Session history, model choice, app settings, extension credentials, provider keys / OAuth, and usage totals live only in this browser (IndexedDB via Dexie).
The app still uses the network: Model requests go directly to the providers you configure, unless you explicitly route them through Settings -> Proxy. Optional extensions may call their own network services when enabled. WebAura does not load third-party analytics or event tracking.
| Setting | What it’s for |
|---|---|
| Settings -> Providers | LLM API keys and OAuth credentials for the model providers you use. |
| Settings -> Extensions | Optional AI-callable browser tools and extension-owned credentials/settings. Disabled extensions are not exposed to the model. |
- Browser workspace - Run local-first AI workflows in the browser without a hosted product account.
- Extension packages - The default experience is plain AI chat; installed packages expose a manifest plus lazy runtime/settings entry points and stay disabled until the user enables them.
- Stack - pi-mono with browser-native state and provider access.
- Local first - Agent work runs in a per-tab
DedicatedWorker; durable state stays on the main thread through IndexedDB. - Resilient - Lease ownership, runtime recovery, and interrupted-turn repair all stay inside the browser runtime.
Inspired by Sitegeist, btca, and repogrep.
This codebase has been built with substantial AI assistance. Very little is hand-written; GPT-5.4 was used heavily to create and iterate on the repository.
WebAura is a fork of gitinspect by Jeremy Osih.
Copyright (C) 2026 Ziphyrien and contributors. Original project: https://github.com/jeremyosih/gitinspect
Licensed under the GNU Affero General Public License v3.0.