RAM —
SYNAPTIC GATE
CORTEX
Local neural stack on your GPU — models, chat, research, and 3D.
CORTEX
Initializing neural interface…
SYNAPTIC GATE
Local neural stack on your GPU — models, chat, research, and 3D.
Global settings for Deep Analysis, Harness, Crew, Code Agent, and OpenCode. Choices are limited by the server allow-list and apply across all projects.
RAM —
—
| Service | Status | Port | Up | RAM | Info | Actions |
|---|
Clippy pipeline
—
Pick a tool from the registry.
Run a tool to see output.
Planner picks appear here.
Context injected into Clippy appears here.
{}
—
SYSTEM PREFERENCES
Configure the Cortex interface.
Choose the language used by the Cortex interface.
LIVE RENDER
Presets and controls. Changes apply instantly.
Default cyan/magenta balance
Variations on fluid, bloom, and wireframe — the pipeline CORTEX already uses.
Tap a blend or tune below
Loading models…
Loading library…
Pick a family, filter good fits, then install a model.
Send a message to test the model
Hold the mic, or type — replies are spoken locally.
Describe a task — the agent writes and runs Python step by step.
Ask something — I will search the web, extract text, embed it, and answer with sources.
Select a project to open the editor. Preview starts in the background — switch tabs anytime.
Loading GitHub…
—
Select a workspace to view its overview.
Select a workspace to view its architecture.
Select a workspace to view its abstraction.
Select a workspace to load its design canvas.
Select a workspace to inspect its database schema.
Start a session to run the coding agent.
Select a table to browse rows.
Run a query to see results.
Select a table.
Select a table.
Select a table to view DDL.
Select a database to view its schema.
Select a database to build query endpoints.
Server tools — process list, variables, engines, schema size.
EXPLORATION
Move the cursor to explore the neural network. The camera follows your gaze.
Multimodal stack
—
Generate sound effects, ambient audio, and music textures with AudioLDM 2. Unloads Ollama from GPU before each run.
Generated WAV appears here (16 kHz).
Download WAVMultimodal stack
—
Text-to-video and image-to-video with Wan 2.2 5B. Unloads Ollama before each run. First generation downloads weights (~10 GB); later runs take several minutes.
Drop a first-frame image, or click to upload
Generated MP4 appears here.
Download MP4Multimodal stack
—
SDXL text→image, Wan 2.2 text/image→video, Ollama vision models, and the Image3D matte → enhance → reconstruct pipeline.
Preview
Output appears here. Upload a source for edit modes — it shows in the corner while you work.
Working…
Upload an image and ask a vision model what it sees.
Drag or click to upload
Reconstruction
Viewport
Add a source image, then generate. Your mesh renders here with orbit controls.
Drag to orbit · scroll to zoom · right-drag to pan
CORTEX · MESSAGING
—
Hit Connect or open Pairing to link your device.
Incoming WhatsApp images and files are archived to .waha-media/ in this project.
—
Scan with WhatsApp → Linked devices
Point your phone at the code below
Live preview appears when the session is connected.
—
—
Welcome to the cortex.