beehiiv
Search for a Claude Code dashboard and you land on Grafana. Community dashboards on grafana.com, an official Grafana Cloud integration, and a setup guide that starts with environment variables.
They work. They are also an observability project. Telemetry switched on, an OTLP endpoint, an access token with write permissions, a stack to receive the data, then a dashboard to import. Worth it for a platform team. A lot for one developer who wants to know whether they will hit the limit before lunch.
What a Grafana Claude Code dashboard needs
Claude Code can export OpenTelemetry. You set CLAUDE_CODE_ENABLE_TELEMETRY=1, choose an exporter with OTEL_METRICS_EXPORTER, and point it at an endpoint. In Grafana Cloud, that endpoint comes from the Claude Code tile under Connections, with an access policy token that can write metrics and logs, built into a Basic auth header. Then you press Install for the dashboards.
Self-hosted, you run the store yourself. The community dashboards on grafana.com read stores such as Prometheus, VictoriaMetrics or Azure Application Insights, and you import the one that matches yours.
None of that is hard for someone who runs Grafana already. It is a lot of plumbing to stand up for one screen.
What Claude Code’s telemetry shows
Claude Code exports eight metrics. Sessions, lines of code, pull requests, commits, cost, tokens, edit decisions and active time. Good material for a chart of adoption or spend across a team, and Anthropic’s docs note that the cost metrics are approximations.
What is not in that list is the number a Pro or Max user actually runs into. How much of the five-hour window and the week is left, and when each one resets. You can watch tokens climb on a Grafana panel and still not know you are at 12%.
One line on your computer
Context Board’s runner is a small program you set up once with one line from Connect. It starts with your computer and reads what Claude Code and Codex last heard from their own servers.
Start from the Agents template, or add the widgets from the gallery. AI usage shows each plan’s five-hour window and week as rings, green at half or more, amber from a fifth, red under that, each with when it resets. Agent activity lists every session on the computer, and Needs you shows the ones waiting on you for a question, an OK or a plan. Git activity shows today’s commits.
From your sessions, only names, states and percentages leave the machine. Prompts and replies never do.
What it costs a solo developer
Grafana is open source, so you can host it yourself. Grafana Cloud’s free tier covers 10,000 active metric series and 50 GB of logs, with 14 days of retention. Pro starts at $19 a month plus usage, with 13 months of metrics.
So on money alone it is close, and Grafana can be free. The cost is the setup, and keeping a pipeline alive for one dashboard. Context Board is $6 a month after a 3-day trial, with nothing to host, and the same board carries your revenue, traffic and to-dos beside your agents.
Where Grafana still fits
Grafana is built for observability across a team or a company. An admin can push Claude Code’s telemetry settings to every developer through managed settings, Grafana Cloud’s integration installs stats, usage breakdown and productivity dashboards, and the numbers sit next to the metrics, logs and traces of everything else you run.
That is a team’s spend and adoption job. Context Board is one person’s board, built around the plan limit and the sessions, and it does not chart token trends. If you run Grafana already, send the telemetry there and keep the board for the limit. If you just want to see your agents, start with the board.
Try it on your own machine
Link your computer with the one line from Connect, start from the Agents template, and work a normal day with the board on a second screen.
Start your Context Board, free for 3 days. The Claude Code usage page shows the widget in detail, and the AI agent dashboard page shows a fuller agents board.