Implementasi Otomatisasi Alert pada Sistem Monitoring Server Menggunakan Prometheus dan AI Agent Berbasis Large Language Model Terintegrasi n8n

Rana Hamida, Dr. Sahirul Alam, S.T., M.Eng.

2026 | Tugas Akhir | D4 TEKNOLOGI JARINGAN

Sistem monitoring berbasis Prometheus tidak menyediakan mekanisme penyimpanan riwayat alert secara otomatis sehingga analisis insiden masih dilakukan secara manual. Penelitian ini mengembangkan sistem otomatisasi pencatatan alert berbasis n8n yang terintegrasi dengan Google Sheets dan AI Agent berbasis Large Language Model untuk analisis interaktif melalui Telegram. Sistem menggunakan dua workflow utama, yaitu pencatatan otomatis alert dan analisis berbasis permintaan pengguna dengan pendekatan Constraint-based Prompting. Hasil pengujian menunjukkan keberhasilan fungsional 100?n akurasi pencatatan 100%. Pengujian AI menghasilkan akurasi deteksi intent dan relevansi jawaban sebesar 100?ngan tingkat halusinasi 0%. Waktu respons sistem berada di bawah threshold, yaitu rata-rata 4,36 detik dan 3,20 detik pada workflow pencatatan otomatis alert serta 13,08 detik pada workflow analisis AI, sehingga sistem dinilai mampu meningkatkan efisiensi monitoring dan analisis insiden secara otomatis.

Prometheus-based monitoring systems do not provide an automatic alert history storage mechanism, causing incident analysis to still be conducted manually. This research develops an alert logging automation system based on n8n integrated with Google Sheets and an AI Agent powered by a Large Language Model for interactive analysis via Telegram. The system utilizes two main workflows: automated alert logging and user-request-based analysis using a Constraint-based Prompting approach. The test results demonstrate 100% functional success and 100% logging accuracy. AI testing achieved 100?curacy in intent detection and response relevance with a 0% hallucination rate. System response times remained below the threshold, averaging 4.36 seconds and 3.20 seconds for the automated alert logging workflows, and 13.08 seconds for the AI analysis workflow. Consequently, the system is considered capable of enhancing monitoring efficiency and automated incident analysis.

Kata Kunci : Workflow automation, sistem monitoring server, Prometheus, Large Language Model.