I used to waste time opening GitHub, checking repos, and manually summarizing what happened each week. It felt like a chore. So I built a small n8n workflow that just pulls the activity, counts it up, and sends a clean digest to Telegram every Monday.
It’s not fancy. It doesn’t use a vector DB or complex AI parsing. It just aggregates the raw counts first, passes a short summary to an LLM to make it readable, and sends it over. If the LLM call fails for some reason, a fallback node still delivers the raw stats. I figured a broken summary is better than no summary at all.
I’ve been running it quietly for a while, and it’s been reliable enough that I finally cleaned up the README, added a quick setup guide, and put it on GitHub.
Quick takeaway:
- Zero dashboard fatigue: Clean weekly summaries land straight in Telegram.
- Token-smart: Aggregates data before calling the LLM to keep costs near zero.
- Never silent: If AI fails, raw stats still get delivered. No silent drop.
- On-demand: Type
/report,/issues, or/statusanytime for instant updates.
🔗 n8n-workflow-github-telegram-digest
I’ve also shared a couple other n8n workflows on my profile - mostly around GitHub integrations and simple bots. Nothing ground-breaking, just practical patterns that solved real headaches for me. Happy to connect if you’re tinkering with similar stuff or just want to swap notes.
Top comments (0)