There are two real, working directions of n8n connectivity today: **n8n pushing content into your
Brain (works from any install) and automatically syncing an Obsidian vault into the Brain via
n8n** (a self-hosted/VPS ops setup, not part of the packaged desktop install). Neither requires the
still-planned MCP control tools covered on n8n — let agents control n8n.
The simplest connection: an n8n HTTP Request node that POSTs to your Brain, the same endpoint
covered in depth on Drive it from any LLM and reachable in-app at
Connect (/connect):
- Get your API key from the Connect page (generates one if you don't have one yet).
- Add an HTTP Request node in your n8n workflow:
POST https://<your-domain>/api/brain/push. - Header:
Authorization: Bearer YOUR_KEY. - Body (JSON):
{ "title": "...", "content": "...", "team": "optional team name", "tags": [...] }. Settingteamroutes the note into that team's folder so the right agents see it first.
This turns any n8n workflow (a scraper, a form submission, a scheduled report) into something your
agent teams automatically pick up.
If you keep your notes in Obsidian and sync them with Syncthing, an n8n workflow can watch for
changed .md files and index them into the Brain automatically every few minutes — this is real
and documented in docs/ops/n8n-vault-indexer.md, requires:
- Syncthing running with the vault populated on the same host as n8n/the app.
- Ollama with
nomic-embed-textpulled (used to embed the notes for semantic search). - A shared
BRAIN_SECRETbetween n8n and the app, andVAULT_CONTEXT_ENABLED=true.
This is infrastructure you set up yourself on a server you control — it's not a toggle in the app
UI, and it isn't part of the one-click desktop install. If you just want to get a note or a piece
of automation output into the Brain, the push endpoint above is simpler and works everywhere.
- n8n — let agents control n8n — Status: planned
- Drive it from any LLM — the same push endpoint, framed for any external tool or LLM, not just n8n