As AI coding agents accelerate development, developers lose the deep understanding they once built by writing code themselves. Telemetry-driven development is proposed as a way to regain confidence: by connecting coding agents (like Claude Code) to real production telemetry via Grafana's gcx CLI and Grafana MCP server, agents can ground their implementations in actual system behavior rather than assumptions. The workflow covers fetching RED metrics to inform new feature specs, updating dashboards automatically, running local OpenTelemetry setups with the grafana/otel-lgtm Docker image, generating realistic load tests with k6 (including AI-authored test scripts), and creating iterative improvement loops using profiling data. The Tempo team at Grafana Labs already uses an agentic harness that baselines performance, finds hotspots, implements changes, redeploys, and compares results automatically. The end goal is PRs backed by dashboard links and telemetry evidence rather than agent-generated descriptions alone.
