npm install @wimoron/llm-cost-guard

npx @wimoron/llm-cost-guard start

import { patch } from '@wimoron/llm-cost-guard';
import OpenAI from 'openai';

const openai = patch(new OpenAI());

// Use openai exactly as before — every call is tracked automatically
const res = await openai.chat.completions.create({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: 'Hello!' }],
});

import { patch } from '@wimoron/llm-cost-guard';
import Anthropic from '@anthropic-ai/sdk';

const anthropic = patch(new Anthropic());

const msg = await anthropic.messages.create({
  model: 'claude-3-5-sonnet-20241022',
  max_tokens: 1024,
  messages: [{ role: 'user', content: 'Hello!' }],
});

curl -X POST http://localhost:47821/api/track \
  -H "Content-Type: application/json" \
  -d '{
    "provider": "openai",
    "model": "gpt-4o",
    "promptTokens": 100,
    "completionTokens": 50,
    "totalTokens": 150,
    "costUSD": 0.00125,
    "latencyMs": 450,
    "userId": "alice"
  }'

import { addBudget } from '@wimoron/llm-cost-guard';

// Block calls when alice spends more than $2/day
addBudget({ scope: 'user', scopeId: 'alice', limitUSD: 2.00, windowHours: 24, hardBlock: true });

// Alert (but don't block) when team spends over $50/day
addBudget({ scope: 'team', scopeId: 'eng', limitUSD: 50.00, windowHours: 24, hardBlock: false });

// Global $200/day soft cap
addBudget({ scope: 'global', scopeId: 'global', limitUSD: 200.00, windowHours: 24, hardBlock: false });

Error: [llm-cost-guard] Budget exceeded for user "alice": $2.0041 of $2.00
  code: 'BUDGET_EXCEEDED'

try {
  const res = await openai.chat.completions.create({ ... });
} catch (err) {
  if (err.code === 'BUDGET_EXCEEDED') {
    return res.status(429).json({ error: 'Daily limit reached. Try again tomorrow.' });
  }
  throw err;
}

npx @wimoron/llm-cost-guard setup

{
  "mcpServers": {
    "llm-cost-guard": {
      "command": "npx",
      "args": ["@wimoron/llm-cost-guard", "mcp"]
    }
  }
}

[mcp_servers.llm-cost-guard]
command = "npx"
args    = ["@wimoron/llm-cost-guard", "mcp"]

npx @wimoron/llm-cost-guard start    # Start dashboard (opens browser automatically)
npx @wimoron/llm-cost-guard setup    # Auto-connect to all detected editors
npx @wimoron/llm-cost-guard status   # Check if server is running
npx @wimoron/llm-cost-guard mcp      # Start MCP mode (used by editors internally)

npm test

import {
  patch,          // patch(client, userId?) — wraps OpenAI or Anthropic
  patchOpenAI,    // explicit OpenAI patch
  patchAnthropic, // explicit Anthropic patch

  addCall,        // manually record a call
  calcCost,       // calcCost(provider, model, promptTokens, completionTokens)

  addBudget,      // addBudget({ scope, scopeId, limitUSD, windowHours, hardBlock })
  removeBudget,   // removeBudget(id)
  checkBudget,    // checkBudget(userId) → { blocked, reason }

  getStats,       // get dashboard stats snapshot
  getCalls,       // getCalls(limit, userId)
  ackAlert,       // ackAlert(id) or ackAlert('__all__')

  startServer,    // start the HTTP server programmatically
} from '@wimoron/llm-cost-guard';