API Cost Tracker

Track AI API costs across OpenAI, Anthropic, Google AI with budget alerts, analytics, and optimization tips

Security Scan

The skill claims live multi-provider cost tracking and alerts, but the included code does not actually call provider APIs or use the advertised API keys/webhooks — the behavior is inconsistent with the description.

Purpose & Capability

The name/description promise real-time tracking across OpenAI, Anthropic, and Google AI using API keys and webhooks. The code, however, only computes costs from provided token counts, stores local demo entries, and does not integrate with provider APIs or fetch usage from those services. The listed optional env vars (OPENAI_API_KEY, ANTHROPIC_API_KEY, GOOGLE_AI_KEY) are plausible for the described purpose but are not consumed by the CLI shown. Webhook alerting is present in config examples but the script contains no network/send logic for webhooks.

Instruction Scope

SKILL.md instructs users to set API keys, run automated tracking, add crontab entries, and configure webhooks. The runtime instructions in scripts/main.mjs (as provided) only create demo tracking entries, save to local data files, and print reports. Several advertised CLI commands (e.g., test-keys, alerts enable, check-budget, rebuild) appear in docs but are not implemented (the main script is truncated and the implemented commands shown are limited). This grants broad expectations to the user that the skill does not meet.

Install Mechanism

No install spec beyond normal npm usage. package.json lists no external dependencies and there are no network download/install steps. This is low install risk.

Credentials

Required env vars are reasonable for the declared purpose (API keys for the three providers) and are marked optional. However, the code shown does not read or use those env vars, so requesting them in docs/config.json is currently misleading. Do not assume keys will be used safely — they are not consumed in the visible code.

Persistence & Privilege

The skill does not request elevated privileges or persistent platform presence (always:false). It writes only to a local ./data directory. Autonomous invocation is allowed by default (normal for skills) but there is no evidence of the skill modifying other skills or system-wide settings.

What to consider before installing

This skill is inconsistent: the docs say it will read provider APIs, send webhook alerts, and use API keys, but the included script only simulates usage (creates demo entries and writes local JSON files). Before installing or providing any API keys: 1) Inspect scripts/main.mjs fully (and confirm whether it actually calls provider APIs or sends webhooks). 2) If you need real cross-provider tracking, ask the author for proof of implemented integrations or an updated release that actually uses the provider APIs securely. 3) Do not paste real API keys into config.json or environment variables unless you verify the code will use them as expected (and that network endpoints are legitimate). 4) If you plan to run automated jobs (cron/heartbeat), test in an isolated environment first. Confidence is medium because parts of the main script were truncated and tests expect exports that may not be present; additional information (full/main.mjs that includes API integration or explicit network code) would raise confidence and could change the verdict to benign if implementation matches documentation.

Comprehensive AI API cost tracking across multiple providers. Monitor spending, set budgets, get alerts, and optimize your AI costs.

Features

• ✅ Multi-Provider Support - OpenAI, Anthropic, Google AI
• ✅ Real-Time Tracking - Monitor costs as they happen
• ✅ Budget Alerts - Get notified when approaching limits
• ✅ Usage Analytics - Detailed insights into API usage
• ✅ Cost Optimization - Tips to reduce spending
• ✅ Export Reports - JSON, CSV, Markdown formats
• ✅ Historical Data - Track costs over time
• ✅ Model Comparison - Compare costs across models

Installation

cd api-cost-tracker
npm install

Quick Start

# Track all providers
node scripts/main.mjs track

# Track specific provider
node scripts/main.mjs track --provider openai

# View analytics
node scripts/main.mjs analytics

# Set budget
node scripts/main.mjs budget set 100 --monthly

# Export report
node scripts/main.mjs export --format markdown --output report.md

Configuration

Edit config.json:

{
  "providers": {
    "openai": {
      "enabled": true,
      "apiKey": "${OPENAI_API_KEY}"
    },
    "anthropic": {
      "enabled": true,
      "apiKey": "${ANTHROPIC_API_KEY}"
    },
    "google": {
      "enabled": true,
      "apiKey": "${GOOGLE_AI_KEY}"
    }
  },
  "budgets": {
    "daily": 10,
    "weekly": 50,
    "monthly": 200
  },
  "alerts": {
    "enabled": true,
    "thresholds": [50, 75, 90, 100],
    "webhook": "https://your-webhook.com/alert"
  },
  "tracking": {
    "autoTrack": true,
    "interval": 300000
  }
}

API Reference

track(options)

Track API usage and costs.

Options:

• provider (string): Specific provider or 'all'
• period (string): 'today', 'week', 'month', 'all'

Returns:

{
  "total": 45.67,
  "providers": {
    "openai": 32.10,
    "anthropic": 10.50,
    "google": 3.07
  },
  "models": {
    "gpt-4": 28.50,
    "claude-3": 10.50
  }
}

analytics(period)

Get detailed analytics.

Period: 'day', 'week', 'month', 'year'

Returns:

• Cost trends
• Usage patterns
• Model efficiency
• Optimization suggestions

budget.set(amount, period)

Set budget limit.

budget.check()

Check current budget status.

export(format, options)

Export cost report.

Formats: 'json', 'csv', 'markdown'

Usage Examples

Track Daily Costs

node scripts/main.mjs track --period today

Output:

💰 API Costs - Today
━━━━━━━━━━━━━━━━━━━━━━━━━━━━
OpenAI
  GPT-4: $12.50 (125K tokens)
  GPT-3.5: $2.30 (230K tokens)
  Subtotal: $14.80

Anthropic
  Claude-3: $8.20 (82K tokens)
  Subtotal: $8.20

━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Total: $23.00
Budget: $200/month (11.5% used)

Set Budget Alert

node scripts/main.mjs budget set 100 --monthly
node scripts/main.mjs alerts enable

Export Monthly Report

node scripts/main.mjs export --format markdown --period month --output monthly-report.md

Compare Models

node scripts/main.mjs compare --models gpt-4,claude-3,gemini-pro

Pricing Reference

OpenAI (per 1K tokens)

• GPT-4: $0.03 (input) / $0.06 (output)
• GPT-4 Turbo: $0.01 / $0.03
• GPT-3.5 Turbo: $0.0005 / $0.0015

Anthropic (per 1K tokens)

• Claude-3 Opus: $0.015 / $0.075
• Claude-3 Sonnet: $0.003 / $0.015
• Claude-3 Haiku: $0.00025 / $0.00125

Google AI (per 1K tokens)

• Gemini Pro: $0.00025 / $0.0005
• Gemini Ultra: $0.0025 / $0.0075

Integration with OpenClaw

Add to your HEARTBEAT.md for automated tracking:

Every 6 hours:
- Run: node /path/to/api-cost-tracker/scripts/main.mjs track
- Alert if budget > 75%

Automated Budget Monitoring

# Add to crontab
0 */6 * * * cd /path/to/api-cost-tracker && node scripts/main.mjs check-budget

Advanced Features

Cost Optimization Tips

Run optimization analysis:

node scripts/main.mjs optimize

Get suggestions like:

• Switch to GPT-3.5 for simple tasks
• Use Claude-3 Haiku for fast responses
• Batch requests to reduce API calls
• Cache common responses

Webhook Integration

Configure alerts to send to webhooks:

{
  "webhooks": [
    {
      "url": "https://your-slack-webhook.com",
      "events": ["budget_exceeded", "high_usage"]
    }
  ]
}

Historical Analysis

# View last 30 days
node scripts/main.mjs history --days 30

# Compare months
node scripts/main.mjs compare --period month --previous

Data Storage

Cost data is stored locally:

data/
├── costs/
│   ├── 2026-03-01.json
│   ├── 2026-03-02.json
│   └── ...
├── budgets.json
└── alerts.log

Troubleshooting

API Key Issues

# Test API keys
node scripts/main.mjs test-keys

Missing Data

# Rebuild database
node scripts/main.mjs rebuild

Testing

npm test

License

MIT - Free for personal and commercial use.

Support

• GitHub Issues: [your-repo]
• OpenClaw Discord: https://discord.gg/clawd

Roadmap

• More providers (Cohere, AI21, etc.)
• Real-time dashboard
• Team cost sharing
• Custom pricing rules
• Predictive analytics