Documentation
How to use LLeMbas Server and LLeMbas CLI, written for the people running and using them.
Start here
- Install the server and create the first account — it becomes the administrator.
- Add a connection to a model endpoint and enable its models.
- Install the CLI on the machine where your code is, and
lembas loginto your server. - Turn on agent chats to run that machine from the browser.
Reading these pages. Menu paths look like Admin → Connections (administrators) or Settings → Devices (everybody). What you see depends on your permissions: if a page in these docs is missing for you, an administrator has not granted it.
Server
- ConnectionsAdd your model endpoints: OpenAI-compatible and Anthropic, provider names, model settings.
- Agent chatsChats in the browser whose work runs on a machine with the LLeMbas CLI.
- LibraryKnowledge bases, notes, memory and skills — what a model can reach for.
- Permissions and sharingGroups, permissions, quotas, read-only sharing and data groups.
- Image generationDrawing with ComfyUI: workflows, defaults and the review pass.
- VoiceDictation and read-aloud against your own speech servers.
- CreditsPrices per model, plans and renewals, balances and the credits API.
- SearchWeb search for models, searching your own things, and semantic search.
- APITokens and scopes, the OpenAI-compatible /v1, MCP, and examples with curl.
- API-only modeRun a server that is an API and nothing else.