n8n — self-hosted automation

Self-hosted n8n instance deployed on Railway, running the scheduled automation pipelines I build and maintain.

This repository holds the Dockerfile I deploy from, so the runtime is reproducible and version-pinned rather than tied to a hosted plan.

What runs on it

Solana token scanner — polls for new token launches on a 30-minute schedule and audits each candidate through RugCheck, Helius, Birdeye and Jupiter: authority checks, liquidity and pool composition, and dev-wallet analysis for bundle sniping. Migration detection is derived from live pool data rather than from documentation — tokens still on the bonding curve are filtered out before anything reaches the alert step. Surviving candidates are pushed to Telegram.

Building it meant handling the unglamorous part of third-party integrations: when RugCheck changed its response shape, the pipeline went silent rather than loud, and finding that took reading actual runs instead of trusting the schema.

Job discovery — collects vacancies from Djinni and getmatch, filters them against a stack profile, and delivers matches to Telegram.

Workflow definitions and credentials are not stored here — they live in the mounted volume, which is why the volume matters more than the image.

Usage

Build the image:

docker build -t n8n-self-hosted .

Run it locally:

docker run -it --rm \
  -p 5678:5678 \
  -v n8n_data:/home/node/.n8n \
  n8n-self-hosted

The editor is then available at http://localhost:5678. The named volume keeps credentials and workflow definitions between restarts — without it, everything is lost when the container is removed.

Notes

• Intended for deployment on a container host (Railway, Fly, a VPS) rather than as a library.
• Stack: n8n, Docker, Railway, Solana data APIs (RugCheck, Helius, Birdeye, Jupiter), LLM APIs, Telegram Bot API.