claude-code - 💡(How to fix) Fix [FEATURE] Add extra usage metric to telemetry [3 comments, 2 participants]

Preflight Checklist

• I have searched existing requests and this feature hasn't been requested yet
• This is a single feature request (not multiple features)

Problem Statement

When managing my team, it would be great to have more granular views of when an how they need extra usage—e.g. which models, projects, or workflows the extra usage is associated with

Proposed Solution

Just like the usage metric, claude code should also report an extra_usage metric that tracks how much extra usage is billed for a given request (above the user's existing subscription plan).

Alternative Solutions

You can see extra usage in the console, but there's no way to graph it or break it down by any sort of detail

Priority

High - Significant impact on productivity

Feature Category

CLI commands and flags

Use Case Example

As an engineering manager with a team of N developers using Claude Code, I need to understand our extra usage patterns to optimize costs and justify budget. Today, I can see a single "extra usage" dollar
amount in the Anthropic console, but I can't answer questions like:

• Which projects drive the most extra usage? Our monorepo team might be generating 80% of extra usage due to large context windows, while the microservices team stays within plan limits. Without per-project breakdown, I can't target optimization efforts.

• Which models are responsible? If most extra usage comes from Opus requests, I could set team guidelines to default to Sonnet for exploratory work and reserve Opus for complex tasks — but I need data to make that case.

• Are there workflow patterns to address? One developer might be hitting extra usage repeatedly on CI debugging loops that could be solved with better test infrastructure. Another might spike during on-call incidents, which is expected and acceptable.

• How do I forecast next month's bill? With a granular extra_usage metric exposed alongside the existing usage metric, I could build a Datadog/Grafana dashboard that shows extra usage trends over time, set alerts when a team or project exceeds a threshold, and produce weekly reports for finance.

Concrete scenario: I export telemetry to our observability stack and create a dashboard with extra_usage broken down by user, project, and model. I notice that our payments team's extra usage tripled last sprint. I dig in and find they were using Opus for a large refactor across 50+ files. That's a legitimate spike — no action needed. But I also spot a bot integration account generating unexpected extra usage from repeated failed tool calls, which I can now fix. Without the extra_usage metric, both of these would be invisible — just a bigger number on the monthly invoice.

Additional Context

No response