Connect an AnythingLLM local server to the Stellary workspace catalog and use compatible models for project missions without separating execution from cards and review.
AnythingLLM
Connected
Stellary
Launch workspace
The local machine owns the server. Stellary Desktop reports readiness and model visibility, while the agent configuration determines where eligible missions run.
01
Run AnythingLLM and verify that its local API target is reachable.
02
Connect the server from Stellary Desktop and refresh discovery.
03
Share only the models intended for the workspace.
04
Choose local execution and the available target on the agent.
05
Run a compatible mission and inspect its return state in Activity.
Mission STL-128
Run AnythingLLM and verify that its local API target is reachable.
Checklist
Linked context
The AnythingLLM server remains on the connected machine and is unavailable when that machine is offline.
Stellary distinguishes local server targets from repository-capable CLIs and only offers models that are ready, visible, and authorized.
The AnythingLLM server remains on the connected machine and is unavailable when that machine is offline.
The user controls which discovered models become selectable by workspace agents.
Choose missions suited to a local response server rather than assuming code or tool access.
1. Start the AnythingLLM local server
2. Connect it in Stellary Desktop
3. Refresh the local model catalog
4. Share the intended target with the workspace
5. Assign the target to a locally executed agentAnythingLLM can join the catalog of local AI apps connected to Stellary Desktop. The workspace stores the usable capability signal, not a copy of the local model itself.
The machine must be online and connected when a mission targets that server. Readiness changes should be visible before the user launches explicit work.
A local API target may be appropriate for drafting, synthesis, or analysis but may not modify a linked repository. This distinction belongs in the product selection flow.
When the card requires a coding loop, use Codex, Claude Code, Cursor Agent, or another compatible CLI runner with the project repository link.
AnythingLLM local server
Agent CLI runner
Interface
AnythingLLM local server
Local API served by AnythingLLM.
Agent CLI runner
Installed command-line agent.
Typical output
AnythingLLM local server
Model response for an eligible mission.
Agent CLI runner
Tool-using work in the linked repository.
Availability
AnythingLLM local server
Server and machine must be ready.
Agent CLI runner
CLI, login, machine, and repository must be ready.
Control
AnythingLLM local server
Workspace sharing and agent policy.
Agent CLI runner
Workspace sharing, repository scope, and tool policy.
No. You run and configure AnythingLLM on your own machine, then connect the available local server through Stellary Desktop.
Only targets you explicitly share with the workspace become available to them. Private discovered models remain visible only to their owner.
No. A mission must match the capabilities reported for that local target. Repository or specialized tool work may require a compatible agent CLI.
Eligible local missions report their result or failure through the Stellary runtime so the team can inspect the state from Activity.
Connect the server, share one ready model, and start with a compatible mission.
Download Stellary Desktop