auditthread

Track AI spend across OpenAI, Anthropic, and Google — in one line.

Install

npm install @frostfrazer/auditthread

Quick start

import { AuditThread } from 'auditthread'
import OpenAI from 'openai'
import Anthropic from '@anthropic-ai/sdk'

const at = new AuditThread({ apiKey: 'at_your_key' })

// ── OpenAI ─────────────────────────────────────────────
const openai = at.wrapOpenAI(new OpenAI({ apiKey: process.env.OPENAI_API_KEY }))

// All calls tracked automatically from here
const res = await openai.chat.completions.create({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: 'Hello' }],
})

// ── Anthropic ──────────────────────────────────────────
const anthropic = at.wrapAnthropic(new Anthropic({ apiKey: process.env.ANTHROPIC_API_KEY }))

const msg = await anthropic.messages.create({
  model: 'claude-sonnet-4-5-20250929',
  max_tokens: 1024,
  messages: [{ role: 'user', content: 'Hello' }],
})

// ── Google Gemini ──────────────────────────────────────
import { GoogleGenerativeAI } from '@google/generative-ai'

const genAI = new GoogleGenerativeAI(process.env.GOOGLE_API_KEY!)
const model = at.wrapGemini(
  genAI.getGenerativeModel({ model: 'gemini-1.5-flash' }),
  'gemini-1.5-flash'
)

// ── Manual tracking (any provider) ────────────────────
await at.track({
  provider: 'mistral',
  model: 'mistral-large-latest',
  prompt_tokens: 500,
  completion_tokens: 200,
})

Options

| Option | Type | Default | Description | |---|---|---|---| | apiKey | string | required | Your at_... key from the dashboard | | baseUrl | string | https://app.auditthread.com/api/spend/ingest | Custom ingest endpoint | | agentId | string | — | Tag all events with an agent name | | mode | "immediate" \| "batch" | "immediate" | Send on every call or queue and flush | | batchIntervalMs | number | 5000 | Flush interval in batch mode | | batchSize | number | 50 | Max events per flush | | debug | boolean | false | Log to console |

Per-agent tagging

// Default agent for all calls
const at = new AuditThread({ apiKey: 'at_...', agentId: 'email-responder' })

// Override per wrapper
const openai = at.wrapOpenAI(new OpenAI(...), { agentId: 'code-assistant' })

Batch mode (high-throughput)

const at = new AuditThread({
  apiKey: 'at_...',
  mode: 'batch',
  batchIntervalMs: 10_000,
  batchSize: 100,
})

// Flush manually before shutdown
await at.flush()
at.destroy()