n8n Workflow: Send Weekly GitHub Digests to Telegram with Qwen

🏆 n8n Verified Creator Workflow
This workflow has been reviewed and approved by the n8n team.
👤 Creator: Do Thanh Vinh (Kevin Do)
🔗 Live demo & docs: vinhautomation.com/en/tools/github-weekly-digest-telegram-qwen
⬇️ Download workflow: n8n.io/workflows/15898

🎯 What It Does

This workflow automatically monitors all GitHub repositories on your account and delivers AI-generated engineering reports directly to Telegram.

• 🗓️ Weekly auto-report: Every Monday at 9 AM
• 💬 On-demand commands: Send /report, /issues, /prs, or /status to your bot for instant updates

Perfect for: Indie hackers, solo developers, open-source maintainers, and small engineering teams who want lightweight weekly visibility without expensive analytics tools or Jira setups.

[GitHub Repos] → [Fetch Events] → [Aggregate Stats] → [AI Summarize] → [Telegram Report] ↓ [On-demand: /report | /issues | /prs | /status]

🚀 Key Highlight: Two triggers in one workflow — weekly schedule for automatic reports, Telegram bot for on-demand queries. Stats are aggregated before being sent to the LLM, significantly reducing token usage. If the AI fails, a fallback code node still formats raw stats — you never lose connection.

✨ Key Features

🔹 Multi-Mode Reports via Telegram

• /report: Full 4-part weekly digest (highlights, per-repo details, insights, priorities)
• /issues: List all open issues with age indicators
• /prs: Show PR status (open/merged/closed)
• /status: Quick health check with emoji indicators (🟢🟡🔴)
• Reports split into multiple Telegram messages → stay within 4096-char limit, mobile-friendly

🔹 Smart Collection & Aggregation

• Pulls complete repo list via GitHub User Repos API → auto-filters archived repos
• For each repo: calls Events API for up to 100 recent events
• Events grouped by repo, counted by type (commits, issues, PRs, releases)
• Only counts (not raw payloads) reach the LLM → significantly reducing token usage

🔹 Comprehensive Error Handling & Fallback

• Switch node handles 3 paths: has activity → generate report | all repos failed → send alert | no activity → silent stop
• If AI model fails → fallback code node formats raw stats without AI
• If all GitHub repos fail to fetch → error alert sent to Telegram
• User always gets feedback — never silent failure

🔹 Token-Efficient & Low Cost

• Repository activity aggregated + compressed into stats summary before LLM call
• No vector DB, no embeddings, no extra subscription
• AI does summarization, not retrieval → minimal token usage per run

Steps:

1. Configure GitHub: Create Fine-grained PAT with Contents Read-only (scoped to repos to monitor) → Add GitHub Header Auth credential in n8n

2. Create Telegram Bot: Create via @BotFather → get bot token + your chat ID (call https://api.telegram.org/bot<YOUR_TOKEN>/getUpdates after sending message to bot)

3. Get OpenRouter API Key: Sign up at openrouter.ai

4. Add Credentials to n8n:

   • GitHub Header Auth (for repo/events API)
   • Telegram Bot API (for send messages)
   • OpenAI API (OpenRouter endpoint)

5. Configure Nodes:

   • "Fetch GitHub Repositories" + "Retrieve GitHub Events": Assign GitHub Header Auth credential
   • "Send Messages to Telegram" + "Send Error Alert": Set your Telegram chat ID
   • "OpenAI Report Model": Assign OpenRouter credential
   • "Telegram Message Trigger": Assign Telegram Bot credential

6. Activate: Enable workflow → Send /report to your Telegram bot to test immediately

🧩 Customization Options

• Filter repos: If you have >100 repos, add pagination handling or filter to specific repos:

  var myRepos = ['owner/repo1', 'owner/repo2'];
  items = items.filter(function(item) {
    return myRepos.indexOf(item.json.full_name) !== -1;
  });

Change schedule: Edit cron in "Weekly Digest Trigger":

• Daily at 9am: 0 9 * * *
• Friday 5pm: 0 17 * * 5
• Twice a week: 0 9 * * 1,4
• Swap the LLM: Change model in "OpenAI Report Model" to any OpenRouter-supported model (GPT-4o, Claude, Gemini) or point to self-hosted model
• Change output channel: Replace "Send Messages to Telegram" with Discord, Slack, Notion, or Email nodes — report format works across all platforms
• Add more commands: Edit "Parse Telegram Command" to accept new commands (e.g., /releases, /contributors) → add logic in LLM prompt under new MODE section
• Single message mode: Remove ||| splitting logic in "Format Telegram Messages" → concatenate all report parts into one message

🔸 PHẦN 6: WHAT'S NEXT TABLE

📈 What's Next? (Extending This Workflow)

This workflow is intentionally simple — simple means cheap, stable, easy to debug. When your use case grows, consider:

Enhancement	When to Add
Social Media Auto-post	Repurpose digest into LinkedIn/Twitter post with Telegram approval button before publishing
Obsidian Sync	Push each report as Markdown to Obsidian vault via Git → searchable project history
Unified Weekly Report	Aggregate GitHub + Gmail + Google Calendar + WakaTime + Notion tasks into single report
Error Knowledge Base	Extract bug patterns from issues → append to JSON KB → feed into RAG chatbot for future Q&A
Multi-channel Delivery	Send same report to Telegram + Email + Discord + Slack simultaneously

⚙️ Requirements Summary

• n8n instance (self-hosted or cloud)
• GitHub Personal Access Token (Fine-grained with Contents Read-only, or Classic with repo scope)
• Telegram Bot token + chat ID
• OpenRouter API key (or any OpenAI-compatible endpoint)
• GitHub account with at least one repository

💡 Cost Note

• ✅ Stats aggregation: Free (runs locally in n8n)
• ⚠️ LLM tokens: Billed by OpenRouter per report generation
• 📊 Token usage is low because only aggregated counts (not raw event payloads) are sent to the LLM
• For accounts with hundreds of repositories, cost per run remains minimal (~$0.001-0.02 depending on model)

Monitor usage via OpenRouter dashboard if running on-demand commands at high frequency.

🤝 Contributing & Feedback

• Found a bug in the anonymized example? → Open an issue
• Have an improvement idea? → Submit a PR
• Want to discuss a similar use case? → Email vinh@vinhautomation.com

📄 License

MIT License — Free to use, modify, and share for learning purposes.
Not for commercial resale without prior agreement.

---

🔗 More by this creator:
github.com/dothanhvinh17 | vinhautomation.com/en/tools | n8n.io/creators/dothanhvinh