AI Spend Observability

TokenGoblin — Catch the little monsters eating your AI spend

Track, analyze, and optimize AI token spending across your autonomous agent workforce. Local-first, privacy-preserving, built for teams running production AI workloads.

Open Dashboard BurnRate — new →

What it does

Track

Token ingestion

Ingest token usage from OpenAI, Anthropic, local models, and custom providers via simple API.

Analyze

Spend intelligence

Break down costs by model, agent, workflow, team, and time. Spot anomalies and drift automatically.

Optimize

Efficiency recommendations

Get actionable suggestions: model routing, prompt compression, caching, batching, and routing rules.

Stack

Cloudflare Workers

API + ingestion + scheduled analytics

Cloudflare D1

SQLite-compatible serverless database

Cloudflare Pages

Dashboard + landing page

Go + TypeScript SDKs

Language-agnostic ingestion clients

Current status

Worker: Deployed (tokengoblin-api)
Dashboard: Pages (tokengoblin-frontend)
Landing: Pages (tokengoblin-landing)
D1 Database: token_goblin (configured)

Next:
- Complete dashboard UI (React + charts)
- Add Stripe billing integration
- Build team/workspace isolation
- Add optimization recommendation engine