RelayPulse monitors LLM proxy services by periodically issuing real API requests with actual token consumption and validating that responses contain meaningful content. It detects ‘fake liveness’ scenarios where HTTP returns 200 but content is empty or erroneous. The system provides visual dashboards with 24h/7d/30d availability heatmaps, supports multiple storage backends including SQLite and PostgreSQL, and enables horizontal scaling via containerized deployment.