@weckr/sdk

AI cost and margin intelligence for SaaS founders.

See exactly which users cost more than they pay, per LLM call, zero added latency.

Try it live

See the dashboard with real data. No signup needed. 👉 https://app.useweckr.com/demo

Install

npm install @weckr/sdk

Usage

import { Weckr } from '@weckr/sdk';
import OpenAI from 'openai';

const openai = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });

const wk = new Weckr({
  apiKey: 'wk_your_key_here',
  plans: {
    free: 0,
    starter: 9,
    pro: 29,
    business: 99,
  },
});

const result = await wk.chat(openai, {
  model: 'gpt-4o',
  messages: [{ role: 'user', content: prompt }],
  userId: user.id,
  feature: 'ai-summary',
  plan: user.plan,
});

See your own data in the dashboard: https://app.useweckr.com/dashboard Try the demo without signing up: https://app.useweckr.com/demo

The original LLM call runs unchanged and returns immediately. After it resolves, Weckr fires an async log to the Weckr API. The log call is fire-and-forget. If it fails or stalls, your request is unaffected.

Get your API key

Sign up at https://useweckr.com.

Supported providers

• OpenAI: gpt-4o, gpt-4o-mini, gpt-4-turbo, gpt-4, gpt-3.5-turbo, o1-preview, o1-mini
• Anthropic: claude-opus-4, claude-sonnet-4, claude-haiku-4-5, claude-3-5-sonnet, claude-3-5-haiku, claude-3-opus
• Gemini: gemini-2.5-pro, gemini-2.5-flash, gemini-1.5-pro, gemini-1.5-flash
• Kimi (Moonshot AI): kimi-k2.6, kimi-k3, kimi-k2.5, kimi-k2 (point the OpenAI client at https://api.moonshot.ai/v1)

Dated variants (gpt-4o-2024-08-06, claude-3-5-sonnet-latest, …) resolve to the matching family by longest-prefix lookup. Unknown models log costUsd = 0 and don't trigger caps.

Caps + downgrades

Set per-plan spending caps in the dashboard. When a user crosses their cap:

• action: 'block' → wk.chat() throws WeckrCapError (LLM call never made)
• action: 'downgrade' → the SDK silently swaps the model for a cheaper one in the same provider (gpt-4o → gpt-4o-mini, claude-opus-4 → claude-sonnet-4, etc.)

Errors come in two flavors:

import { isWeckrCapError, isWeckrConfigError } from '@weckr/sdk';

try {
  await wk.chat(openai, opts);
} catch (err) {
  if (isWeckrCapError(err))   return showUpgradePrompt(err);
  if (isWeckrConfigError(err)) return logBackendAlert(err); // typo'd api key, unknown plan
  throw err; // real LLM error
}

Short-lived processes (Lambda, cron, CLI)

wk.chat() returns as soon as the LLM call resolves; the log POST is fire-and-forget. In short-lived processes call await wk.flush() before exit to give the POSTs time to land:

await wk.chat(openai, opts);
await wk.flush(); // default 5s timeout

What gets logged

{
  userId, feature, model, provider,
  inputTokens, outputTokens,
  costUsd, latencyMs,
  planName, planRevenueUsd, marginUsd,
  timestamp,
}

Cost is computed from public per-token pricing. Margin is planRevenueUsd - costUsd (negative means you're losing money on that user).

Dashboard

View cost and margin data at https://useweckr.com/dashboard.

License

MIT