Claude Code Gains Per-Category Usage Breakdown in VS Code
Claude CodeClaude Code v2.1.174 adds a granular usage attribution panel to the Account & usage dialog in VS Code, breaking down consumption over 24h and 7d by cache misses, long context, and subagents, plus per-skill, agent, plugin, and MCP server. It gives VS Code users diagnostic data to understand which workflow components are burning through rate limits fastest.
A diagnostic view into where your usage goes
Claude Code v2.1.174 adds a per-category usage breakdown to the Account & usage dialog inside the VS Code extension. Instead of a single opaque consumption figure, users now see exactly which parts of their workflow are driving usage.
What the panel breaks down
The panel attributes consumption across both a 24-hour and a 7-day window, split by category:
- Cache misses — how much usage is going to prompt segments that could not be served from the cache.
- Long context — the cost attributable to large context windows.
- Subagents — usage consumed by spawned agents rather than the main session.
Beyond those categories, the breakdown also attributes usage per individual skill, agent, plugin, and MCP server, so it is possible to see, for example, that a particular MCP server or a single skill is responsible for an outsized share of consumption.
Why it matters
Rate limits are one of the most common sources of friction for heavy Claude Code users, and until now there was little visibility into what was actually consuming the budget. This breakdown turns that guesswork into data: a developer who keeps hitting limits can open the dialog, identify the cache-miss-heavy or long-context-heavy component, and adjust their workflow accordingly. It is a practical, everyday diagnostic that any VS Code user can act on without leaving the editor.