Msty Nexus One catalog
Browse local runtimes and hosted provider models in one list. Filter, hide, pin, and alias the models your apps should use.
Msty Nexus
One local control center for your AI stack.
Msty Nexus brings local runtimes, provider models, presets, aliases, and secure gateway access into one app so Studio, Claw, scripts, and compatible clients can use the same stack.
Local gatewayProvider keys stay in NexusExplicit LAN access
One gateway
Expose local OpenAI-style and Anthropic-style gateway surfaces that Studio, Claw, scripts, and compatible clients can share.
One key vault
Keep provider credentials inside Nexus and issue scoped local tokens to the apps that need access.
One approval flow
Review app name, callback, scope, and verification details before granting client access to your local stack.
A tour of the app
Four surfaces. One quiet console.
Nexus is designed to remove duplicate model setup, scattered credentials, and one-off runtime configuration from your AI workflow.
Providers
Provider control without the spreadsheet.
Add hosted providers, sync models, hide providers temporarily, and keep credentials behind the local Nexus boundary.
Models
Every model in one catalog.
List Ollama models, llama.cpp builds, MLX checkpoints, Anthropic, OpenAI, OpenRouter, and compatible APIs together.
Presets
Reusable presets, not copy-pasted prompts.
Bundle model settings, parameters, and routing preferences into presets that connected apps can reference by name.
Usage Analytics
See token usage across your stack.
Track provider-reported token totals, request counts, failures, and usage trends by provider, model, and client token from one local console.
Credentials & approval
Provider keys stay in Nexus.
Enter a provider credential once. Studio, Claw, scripts, and shell tools authenticate against Nexus with local tokens instead of upstream provider keys.
- Write-only credential intake. Keys can be set or rotated, but not read back from the app surface.
- Local Nexus access tokens. Create scoped tokens for apps and scripts, then revoke them from one place.
- Per-app approval flow. Review the app name, callback URL, scope, and verification code before granting access.
- Explicit network exposure. LAN access, CORS, and bind-address settings are deliberate choices, not defaults.
The gateway
One endpoint your apps can trust.
Nexus resolves the model or preset, routes each request to the right local runtime or upstream provider, applies the credential, and streams back the response.
Client apps
Msty Studio
local token · chat, models
Msty Claw
local token · agents
Custom script
base URL · Nexus token
Nexus Gateway
POST /v1/chat/completions
POST /v1/responses
POST /v1/messages
GET /v1/models
GET /nexus/services
GET /nexus/presets
OpenAI + AnthropicToken Auth
Upstream
Ollama
local runtime
llama.cpp
local runtime
Anthropic / OpenAI
provider APIs
Local runtimes
Local AI without the babysitting.
Nexus knows how to install, version-pin, start, stop, and watch the runtimes that power local inference.
Ollama
Install, pull, run, and expose mainstream local models through a shared catalog.
llama.cpp
Bring your own GGUFs and manage raw inference services with clearer defaults.
MLX
Use Apple silicon native runtime support where Nexus can manage model records and service state.
Network access
Control what can connect.
What can talk to it
OpenAI-style. Anthropic-style. One local gateway.
Connect first-party Msty apps, SDK-based tools, shell workflows, or your own scripts when they support configurable API endpoints and local tokens.
01
Msty Studio
02
Msty Claw
03
OpenAI SDK
04
Anthropic SDK
05
curl / shell
06
Custom scripts
Questions
Asked & answered.
Is Nexus a hosted service?
No. Nexus is software you run on your own machine or a server you control. The gateway it exposes is a local endpoint that apps and scripts on the same machine, or on your LAN if you choose, can use.
Where do provider API keys go?
Provider credentials stay in Nexus. Client apps authenticate with local Nexus tokens instead of receiving the upstream provider keys.
Is Nexus open source?
Much of Nexus will be open source so teams can inspect, run, and extend the local gateway with confidence. Some Teams and Enterprise features will remain commercial.
Does this mean nothing leaves my machine?
Not if you use a cloud provider. Cloud-provider calls still go to that provider. Nexus controls routing and credential use from your local runtime.
Which apps can connect?
Msty Studio and Msty Claw are first-party targets. Other clients can connect when they support OpenAI-style or Anthropic-style API configuration.
What about LAN access?
LAN access is off by default. You turn it on explicitly and configure CORS and bind-address behavior from one place.
Get Nexus
One quiet app for the messy middle.
Manage local services, providers, models, presets, and app access from one place.