For individuals getting started
For serious Claude Code users
For teams and shared budgets
Everything in Pro, plus:
We don't charge per token or per request.
You pay for predictable alerts — not usage spikes.
CostFence uses Claude Code's official OpenTelemetry (OTEL) metrics exporter. There are no SDKs, wrappers, or code changes required — just enable the exporter and point it to CostFence.
CostFence is not affiliated with Anthropic, but it relies on documented and supported Claude Code metrics. If Claude Code emits it, CostFence can alert on it.
Alerts typically fire within seconds of a limit being crossed. This means you can stop a runaway session before it drains your budget.
No. CostFence never stores prompts, responses, or source code. We only ingest usage metadata (tokens, cost, session IDs, timestamps).
Nothing breaks. Claude Code continues to run normally. You simply won't receive alerts during the outage. CostFence is out-of-band monitoring, not a dependency.
Yes — depending on your plan.
We focus on signal, not notification spam.
Because alerts shouldn't get more expensive when things go wrong. CostFence pricing is based on predictable alert value, not usage volume.
Those tools are great for analysis after the fact. CostFence is built for interrupting problems while they're happening. If you want charts, use a dashboard. If you want to stop surprise bills, use CostFence.
Yes. The Team plan supports:
Today, CostFence is laser-focused on Claude Code.
That focus is what allows us to deliver faster setup, better alerts, and simpler pricing.
We may expand later — but not at the cost of simplicity.
Yes — the Free plan is unlimited time, no credit card required. Upgrade only when you need more alerts or team features.
Most users are up and running in under 5 minutes. Enable OTEL → set your limits → wait for your first alert.
If you want deep analytics, model comparisons, or long-term reporting — CostFence may not be the right tool. If you want instant visibility into Claude Code costs, it is.