Skip to content

LM Studio Project Agents with Stellary Desktop

Use models served by LM Studio as local Stellary execution targets while the project, cards, documents, assignments, and results remain in the shared workspace.

Connect LM StudioOpen the local AI guide
LM Studio · local API server

LM Studio

Connected

Stellary

Launch workspace

Ready
Project context
Prepare release mission
Activity
Identity, project, and tools remain scoped

Make a local model available to the workspace

Stellary Desktop connects to the local server, exposes ready models, and lets you decide which models other workspace agents may use.

  1. 01

    Serve

    Start LM Studio and make the selected model available through its local API.

  2. 02

    Connect

    Add or detect the local server from Stellary Desktop.

  3. 03

    Share

    Expose only the ready models authorized for the current workspace.

  4. 04

    Assign

    Select local execution and the compatible LM Studio target on an agent.

  5. 05

    Run

    Launch an eligible mission and keep its result in Activity.

Mission STL-128

Start LM Studio and make the selected model available through its local API.

Running

Checklist

Connect
Share
Assign

Linked context

project-brief.md

Only models reported as ready by a connected machine appear as usable targets.

Run mission

Local inference, shared project control

The model runs on the connected machine. Stellary controls whether it is visible to the workspace and keeps the mission attached to the project.

Local model catalog

Only models reported as ready by a connected machine appear as usable targets.

Controlled sharing

Private local models stay private until the user explicitly shares them with the workspace.

Honest capability boundary

Local server responses are useful, but some targets do not edit repositories or support every agent tool.

Local server checklist

Text
1. Start LM Studio local server
2. Open Stellary Desktop
3. Connect or refresh local AI apps
4. Share the ready model with the workspace
5. Select it on an agent using local execution

Use LM Studio when local inference matters

LM Studio is useful when a team wants a model served from its own machine for project responses. The machine must remain connected and the server must report the model as ready.

Model privacy and workspace availability are separate choices: seeing a model yourself does not automatically make it available to every member or agent.

Match the mission to the runner capability

A local API server may produce text without offering the repository-editing behavior of an agent CLI. Stellary should expose that distinction instead of presenting every model as interchangeable.

Use LM Studio for eligible response or analysis missions. Use a CLI runner such as Codex or Claude Code when the mission needs to operate directly in a linked codebase.

Local server or coding CLI?

Primary capability

LM Studio local server

Serve model responses through a local API.

Agent CLI runner

Operate as a coding agent in a repository.

Repository changes

LM Studio local server

Not assumed.

Agent CLI runner

Available when the runner and policy support them.

Model choice

LM Studio local server

Models loaded and served locally.

Agent CLI runner

Models exposed by the installed CLI.

Best fit

LM Studio local server

Private local inference and response tasks.

Agent CLI runner

Tool-using code missions.

Questions about LM Studio and Stellary

Does Stellary run LM Studio in the cloud?

No. LM Studio runs on the connected local machine. Stellary Desktop reports the available server and models to the workspace catalog.

Are all local models shared automatically?

No. The user chooses which ready models are shared with the workspace. Other local models remain private.

Can an LM Studio model edit my repository?

Repository editing is not assumed for local API servers. Use a compatible agent CLI runner when the mission needs direct codebase operations.

What happens when the machine disconnects?

Its local targets are no longer available for new missions. The interface should show that state instead of allowing a silent failure.

Continue the setup

Local AI apps guideConnect servers, share models, and configure agent targets.AI agent project managementChoose the role, context, and autonomy around the local model.Claude Code integrationUse a CLI runner when repository work is required.

Connect one local model to a real project

Keep inference on your machine while the mission and review remain in Stellary.

Download Stellary Desktop