Cortex Atlas
A local and cloud AI chat WebUI with guarded tool access and no runtime dependencies.
The project
One chat surface for the models you choose.
Cortex Atlas is a provider-agnostic chat WebUI that runs locally and talks to the providers you configure — Ollama, OpenRouter, OpenAI, DeepSeek, or a custom endpoint — through normalized provider adapters and server-sent events.
It is built as plain Node.js and vanilla JavaScript with zero runtime dependencies, and its guarded tool access is designed so requests leave the machine only toward the providers in use.
What it explores
- Local and cloud provider adapters behind one chat UI
- Streaming responses over server-sent events
- Guarded tool access
- Zero-dependency Node runtime with doctor and verification scripts
Where it stands
Version 0.6.1 is recorded in the local manifest, with verification, acceptance, and conformance scripts in the tree. No public release is confirmed.
In the ecosystem
A hands-on AI surface alongside the other assistant projects. No integration with them is claimed.
Keep exploring
Nearby in the ecosystem.
AnxCorTex HQ
A desktop headquarters for AI missions, agents, and approval-driven work.
Project ECHO
A local-first voice assistant for a more personal computing environment.
ANX Model Lab
Exploring native AI models, from the first weights to evaluation.