Tab · Insights
Claude Code usage
See exactly what your Claude Code sessions cost — and where to spend less without losing speed. Per-session token spend rolled up from the local Claude Code journal, with a heuristic plus LLM recommendation for what to switch.
See it in motion
Where to find it
- Localhost:
/claude-code-usage.html?repo=<id> - API:
GET /api/cc-telemetry?repo=<id> - Keyboard: ⌘ K then
usage - Sidebar: Insights → Claude Code usage
What it does for you
Per-session cost, no spreadsheets.Each row is one session — tokens in, tokens out, cost in USD, tier (Opus / Sonnet / Haiku). Sort by cost to find the expensive day; sort by tokens-out to find the runaway loop.
Heuristic + LLM recommendation, side by side.Every row carries a cheaper-tier hint when the prompt class would have run fine on Sonnet or Haiku. Accept the recommendation or override it from the row.
The rec lands as a PR, not a dashboard toggle.When a recommendation is worth applying repo-wide, the routing control tab opens a PR against
llm-routing.json with the change.Configure
Nothing — telemetry comes from ~/.claude/projects/<repo>/sessions/*.jsonl on the developer's machine. The aggregator reads it directly.
Use it well
Open this tab on Monday morning. Skim the weekend column — the cost-per-session distribution is usually bimodal: a long tail of small queries plus a few expensive long-context runs. The expensive ones are where switching tiers actually moves the bill.