Official documentation: https://docs.n8n.io/
-
Copy and fill the
template.env
. Rename it tolocal.env
to provide environment variables (see:<TODO>
). Passwords for n8n can be generated securely viaopenssl rand -base64 32
-
Run n8n-docker-compose.yml to start up the n8n self-hosted service:
podman-compose -f docker-compose.yml up -d # if you have a docker license: docker compose -f docker/docker-compose.yml up -d
-
Open http://localhost:5678/ and create a fake owner account
-
Create a credential for iteraGPT API https://docs.n8n.io/credentials/add-edit-credentials/ with a unique API Key.
-
Start building agents 🤖
-
Redis
Redis is included in the Docker Compose setup. Snapshots are saved automatically (REDIS_ARGS=--save 60 1000):
- every 60 seconds,
- but only if at least 1000 keys have changed in that time
-
Qdrant
- Currently runs without API key
-
Workflows
To use the provided JSON workflows you need to import them manually and re-connect subworkflows:
- Create a new workflow in n8n.
- Rename the workflow so its name matches the JSON file you want to import
(e.g. if the file isprojekt_referenz_schreiber.json
, rename the workflow toprojekt_referenz_schreiber
). - Import the JSON file into the newly created workflow.
- After all JSONs are uploaded, you must reapply the subworkflows for
projekt_referenz_interview
:
- Open the node that calls a subworkflow.
- Select the workflow that matches the node’s name.
For example, inprojekt_referenz_interview
there is a node callingprojekt_referenz_schreiber
.
You need to open that node and select the workflowprojekt_referenz_schreiber
manually,
since this is not done automatically by n8n.
➡️ In the future this process could be automated via the CLI (
n8n import:workflow
), but for now the manual steps above are required.- Alternative: CLI import (see n8n import docs).
-
Community Nodes
- The Community Node "n8n-nodes-qdrant" may be required
- See Community Nodes in n8n for installation instructions