Self-host the Docker server & its developer API

USE 0 - 20 min

Run the multi-user server and call its developer API

The desktop app lacks multi-user support; the Docker server has it (official feature comparison at docs.anythingllm.com/installation-docker/overview lists "Multi-user support" as ✅ Docker, ❌ Desktop). For a shared lab endpoint, deploy the mintplexlabs/anythingllm Docker image on a spare machine so the whole team logs in and chats with one private knowledge base. The Docker server also exposes a full developer API — "AnythingLLM supports a full developer API that you can use to manage, update, embed, and even chat with your workspaces" (docs.anythingllm.com/features/api) — with all endpoints browsable at /api/docs on your running instance.

1 Run the Docker server. On a machine with Docker installed, pull and run the official mintplexlabs/anythingllm image (see docs.anythingllm.com for the current run command and volume/port mapping). Open the web UI it serves in your browser.
2 Complete first-run setup in the browser: pick your LLM provider (local via Ollama / LM Studio, or a cloud key) and create a workspace, exactly as you did in the desktop app.
3 Add users so a team shares one private workspace over the same documents, each with their own login — this is the multi-user capability the desktop app lacks.
4 Generate an API key. Accounts with the correct access level can create and delete API keys on the fly (docs.anythingllm.com/features/api). Create one now.
5 Browse your instance's API reference at /api/docs and make a test chat request to a workspace using your key. Confirm you get a document-grounded answer back via the API.

✓

The AnythingLLM Docker server is serving a web UI you can log into, you added a second user, and you got a workspace-grounded response via the developer API using a generated key.

BUILD 20 - 30 min

Call your private workspace from a script via the developer API

The developer API lets you drive AnythingLLM programmatically — send a question to a workspace and get a document-grounded answer back without opening the browser. Explore the available endpoints at /api/docs on your running instance.

Your task

Write a short curl command (or a few lines of Python using `requests`) that authenticates with your API key and sends a question to a workspace, then confirm the response is grounded in your documents.

1 Open /api/docs on your running AnythingLLM Docker instance and find the workspace-chat endpoint.
2 Write a curl command (or minimal Python requests script) that sets your API key in the request header and sends a question to the workspace.
3 Run it and check the response includes an answer drawn from your workspace documents.
4 Note the endpoint path and request shape — this is how you could automate document Q&A from any tool that can make an HTTP request.

Deliverable

A working curl command or short script that hits your self-hosted AnythingLLM developer API and returns a document-grounded answer from a workspace.