Xavier

Local AI Command Center

Conversation Interface

Chat Core

Ready
Voice: browser fallback

Multi-Agent Project Captain

Project Captain

Route ideas, generate Claude handoffs, and review Claude output before Codex-style verification touches real files.

Route An Idea

Captain route will show here.

Review Claude Output

Claude review will show here.

Saved Project Decisions

Saved decisions will show here.

Ecosystem Project Board

Board items will show here.

Captain Map

Open this page to load Xavier's project-captain map.

Feature Audit

Capability Dashboard

This is Xavier's working map: what is solid, what is partial, what is experimental, and what should be improved next.

Camera + Audio Fusion

Vision Core

Turn on your webcam with browser permission, capture frames, run local vision-model-ready analysis, and fuse what Xavier hears with what he sees.

Camera off - press Start webcam
Vision idle

Vision Agent Proof

This panel shows the three-agent flow: encoder -> audio/visual fusion -> response decision.

No frame analyzed yet.

Superpowers Core

Always-on background agents that enforce brainstorming, planning, TDD, code review, integration checks, and pinned local core memory every session.

Pin a Core Memory

Use this for long-term Xavier project rules you do not want to re-explain. It stays local in SQLite.


      

Reflection Engine

After every conversation, Xavier reflects, extracts important pinpoints, connects them to Memory + Teacher agents/Core Memory/Knowledge Library, and makes them fast to retrieve later.

Reflection Engine will populate after chats.

Self-Teaching Core

Auto-start agents that catalog trusted source packs, sort them into Knowledge Library shelves, distill lessons, route lessons into teacher agents, and use Reflection + Core Memory to decide what Xavier should study next.

Self-Teaching Core starts automatically when Xavier starts.

Runtime Self-Knowledge

This is the non-secret self-knowledge packet Xavier receives so he knows what model/provider, memory, learning, voice, vision, and Thought Zone systems are active instead of guessing.

Runtime status has not loaded yet.

Integration Audit

A full cross-check that verifies every Xavier subsystem is present, wired, and separated correctly: chat, voice, vision, Thought Zone, code teachers, 3D Code Lab, computer-control scaffold, Superpowers Core, Core Memory, backups, diagnostics, and guardrails.

Open this page to run the audit.

3D Local Memory Brain

Thought Zone

Drag to spin. Scroll to zoom. Chats, AI replies, uploads, notes, tasks, and memories become nodes. Local agents connect what belongs together.

Thought Zone Status

Graph waiting to load.
Thought AI Reply Memory Knowledge Note Task
Hover or click a node to inspect it.
Connection reasons will appear here after selecting a node or running Review Connections.

Upload Knowledge

TXT, MD, PDF, DOCX, CSV, code files, and JSON can be indexed into local memory search.

Knowledge Library

A local source catalog for encyclopedias, dictionaries, etymology, WordNet-style meaning graphs, and language examples. This page shows what folders are ready; full dump ingestion is staged so Xavier stays local and controlled.

Language Teacher Agents

These background agents teach Xavier grammar, speaking flow, meaning, tone, rhetoric, and multilingual awareness without dumping raw encyclopedias into core memory.


      

Code Teacher Agents

Coding docs stay on the Code Knowledge Library shelf. Teacher agents distill lessons for Python, JavaScript, TypeScript, HTML/CSS, SQL, PowerShell, Rust, Go, C#, Java, and more.


      

Watch Xavier Build

Xavier's Workshop

Tell Xavier what to build in Chat Core. He inspects the files, drafts it, and tests it on his own in the background - this page just shows you what he's doing and lets you preview the result.

No active build yet.Ask Xavier to build something in Chat Core, then come back here.
Nothing to preview yet. Ask Xavier to build something in Chat Core.

Recent Builds

Recent builds will show here.
What changed
No changes to show yet.

Computer Control Scaffold

This is the locked foundation for future PC control. It plans actions, creates approvals, and keeps dangerous controls blocked until allowlists/rollback are built.


      

Memories

Notes / Tasks

Coding Task Queue

Autonomous Coding Worker

Coding Failure Lessons

Approvals

Phone + Private Network

Remote Access Readiness

Shows whether Xavier is ready for your phone through Tailscale and whether Aider is enabled for draft-only coding work.


      

Health/System Status

Diagnostics/Errors

Backups

Audit Logs

Settings

Edit .env for provider settings. Runtime saved settings appear here. Secret values are masked.