Archestra-Prometheus integration

Archestra natively exposes Prometheus metrics for comprehensive monitoring of AI agent infrastructure, enabling organizations to track LLM usage, costs, and security events.

Metrics exported

• llm_tokens_total — total tokens consumed across all providers
• llm_request_duration_seconds — LLM API call latency
• http_request_duration_seconds — gateway HTTP request timing
• Tool guardrail blocking events
• Per-team, per-agent cost attribution metrics

Use cases

• Cost monitoring and budget alerting
• Performance optimization
• Security incident detection
• Capacity planning for AI infrastructure

Configuration

Metrics are available at /metrics endpoint. Configure Prometheus scrape jobs to collect data.

Tradeoffs

• Metrics storage costs scale with cluster size
• Requires Prometheus expertise for effective alerting