Master MCP Server

# Performance Troubleshooting ## Symptoms - Elevated latency on tool calls - Increased error rates or timeouts ## Checks - Inspect `/metrics` (Node) and platform dashboards (Workers, Koyeb) - Verify backends’ `/health` and logs - Confirm load-balancing strategy is appropriate for your topology ## Tuning - Increase `retry.maxRetries` and `baseDelayMs` judiciously - Switch to `health` strategy and feed health scores when available - Tighten circuit breaker thresholds to fail fast on unhealthy instances ## Environment - Run Node with adequate CPU/memory; consider horizontal scaling - Use KV-backed tokens in Workers to reduce token cache misses