See the real drivers
Break down input, cache, output, model, effort, provider, session, and subagent usage.
Private by design · runs locally
Find what is burning your AI budget.
Turn local Claude Code and Codex history into useful answers: which session spiked, which model drove it, what subagents multiplied the context, and when to split, summarize, or step down.
No account. No telemetry. No prompt text in the dashboard payload.
2providers: Claude + Codex
GPT-6current OpenAI family
Claude 5.5current model pricing
100%local analysis
Usage intelligence, not another invoice.
Costs are API-equivalent estimates. Subscription plans use published message windows and fair-use limits, so AI Spend keeps plan pressure, API dollars, and OpenAI credits separate.
Price modern work
Includes GPT-6 Astra, GPT-6.1 Sol and Terra, GPT-6 Luna, and Claude Fable, Opus, Sonnet, and Haiku tiers.
Know what to change
Spot context-heavy threads, fan-out, burst pressure, cache reuse, and the turns worth restarting.
Three commands
Run it on your machine.
Node.js 22+ is recommended. The dashboard binds to localhost by default.
npm install
npm test
npm run dashboard
# open http://127.0.0.1:9020/