Hi everyone,
I’ve published a community node for the LiteLLM proxy management API: @motaouakel/n8n-nodes-litellm
Why this and not the OpenAI node
n8n can already send completions to a LiteLLM proxy — the OpenAI node and OpenAI Chat Model sub-node both accept a custom base URL. This node deliberately doesn’t touch /chat/completions. It covers the part that was still manual: who gets a key, what they’re allowed to spend, and what they actually spent.
What’s in it — 25 operations across 5 resources
- Key — Create, Get, Get Many, Update, Delete, Block, Unblock
- Team — Create, Get, Get Many, Update, Delete, Add Member, Remove Member
- User — Create, Get, Get Many, Update, Delete
- Model — Get Many, Get Group Info, Get Health
- Spend — Get Logs, Get Report*, Get Tags
Model and team pickers load live from your proxy, so you choose from a dropdown rather than typing IDs. The node also works as a tool for AI Agent nodes.
* Get Report hits /global/spend/report, which LiteLLM gates behind an Enterprise licence. Everything else works on the OSS build.
Install (self-hosted): Settings → Community Nodes → Install → @motaouakel/n8n-nodes-litellm
Credentials are just your proxy’s base URL and an admin or master key.
Notes
- MIT licensed, no runtime dependencies, published from GitHub Actions with npm provenance.
- Use
http://127.0.0.1:4000rather thanlocalhostif your proxy runs in Docker on the same host — n8n resolveslocalhostto IPv6 and Docker usually publishes IPv4 only. - Submitted for verification; automated review has passed, manual review is in progress. Until that completes it’s self-hosted only.
npm: https://www.npmjs.com/package/@motaouakel/n8n-nodes-litellm
GitHub: https://github.com/motawakel/n8n-nodes-litellm
Tested against LiteLLM main-latest and n8n 2.8.4. Issues and suggestions welcome on GitHub.