ai-tokens

AI Token Enforcement (Local)

The pattern most AI companies need — a hard budget gate before the LLM call, soft metering after for accurate billing.

What this shows:

• Hard limit as a pre-call gate check (rough estimate, blocks if over budget)
• Soft limits for actual input/output token billing (records overages)
• Per-model credit definitions with overhead cost and pricing
• Simulated LLM calls with accumulated token tracking

Run it:

npm install
npx tsx index.ts

No Limitr Cloud account needed. Everything runs locally.