The autonomous cost monitoring skill for OpenClaw agents. Cost Guard watches your AI agent's API spend in real-time, catches runaway costs before they hit your wallet, and delivers daily spend reports — all without human intervention.
Built by Lemnos AI, the operations-first AI company.
Every OpenClaw operator has the same nightmare: you deploy an agent, go to sleep, and wake up to a $200 API bill from a single hallucination loop. Cost Guard kills that problem.
It runs silently alongside your agent, monitors token consumption across providers, and triggers alerts + automatic throttling when spend exceeds your thresholds. No dashboards to check. No spreadsheets to update. It just works.
| Metric | Value |
|---|---|
| Organic Downloads | 253 in first 14 days |
| Active Deployments | Growing daily across US, EU, and Asia |
| ClawHub Rating | Published v1.1.0 |
| GitHub | Open source, MIT licensed |
| Languages | English + 中文 (Mandarin) |
"If DeepSeek marked a milestone for open-source LLMs, then OpenClaw represents a similar turning point for open-source agents." — Counterpoint Research
🔍 Real-Time Spend Tracking — Monitors Anthropic API costs per session, per day, and per task. No manual logging.
🚨 Threshold Alerts — Set daily/weekly/monthly spend caps. Get notified via your agent's messaging channel (Telegram, Slack, Discord) the moment you approach a limit.
🛑 Auto-Throttle — When spend hits your hard cap, Cost Guard gracefully throttles agent activity instead of letting it burn cash on repeat failures.
📊 Daily Spend Reports — Automatic end-of-day summaries: total spend, cost per task, model breakdown, anomaly flags.
🌐 Multi-Provider Ready — Architected for Anthropic today, extensible to DeepSeek, OpenAI, and domestic Chinese LLM providers (Qwen, Doubao, Kimi) in v2.
📱 Zero-Config Deployment — Install as an OpenClaw skill. No API keys to configure beyond what your agent already uses. Running in under 60 seconds.
# Install via ClawHub (recommended)
openclaw skill install cost-guard
# Or clone from GitHub
git clone https://github.com/getlemnos32/cost-guard.git
cd cost-guardEdit config.yaml in the skill directory:
# Cost Guard Configuration
monitoring:
provider: anthropic # Primary provider to monitor
daily_budget: 15.00 # Daily spend cap in USD
alert_threshold: 0.80 # Alert at 80% of budget
hard_cap: true # Auto-throttle at 100%
alerts:
channel: telegram # telegram | slack | discord
frequency: realtime # realtime | hourly | daily
reporting:
daily_summary: true # End-of-day spend report
weekly_rollup: true # Weekly trend analysis
cost_per_task: true # Per-task cost attributionopenclaw skill status cost-guard
# Output: ✅ Cost Guard v1.1.0 — Monitoring active — $0.00 today┌─────────────────────────────────────────────────┐
│ YOUR OPENCLAW AGENT │
│ │
│ Task Request ──► LLM API Call ──► Response │
│ │
│ │ │
│ │ │
└────────────────────────┼────────────────────────┘
│
┌──────────▼──────────┐
│ │
│ COST GUARD │
│ │
│ • Token counter │
│ • Spend calculator │
│ • Threshold check │
│ • Alert dispatcher │
│ • Report generator │
└──────────┬──────────┘
│
┌──────────▼──────────┐
│ IF over budget: │
│ → Alert owner │
│ → Throttle agent │
│ → Log anomaly │
└─────────────────────┘
We run Cost Guard on our own production agents at Lemnos AI. Here's what it looks like in practice:
- Daily agent budget: $15.00
- Actual daily spend: $0.48 average (97% under budget)
- Alerts triggered: 0 hard-cap events in 30+ days
- Cost per lead generated: < $0.03
- Agent uptime: 24/7, autonomous, zero human monitoring
Cost Guard isn't a concept — it's running in production, protecting real revenue-generating agents, every day.
Operators first. AI developers second.
Lemnos AI builds autonomous agents that replace $80K/year administrative overhead for $12K/year. We specialize in property management, real estate operations, and B2B workflow automation on the OpenClaw stack.
Cost Guard is our first open-source skill — a proof point that AI operations tooling should be free, reliable, and built by people who actually run agents in production.
🌐 getlemnos.ai · 🦞 ClawHub · 𝕏 @GetLemnosAI · LinkedIn
- v1.0 — Anthropic API spend tracking + daily reports
- v1.1 — Threshold alerts + auto-throttle + Mandarin README
- v2.0 — Multi-provider support (DeepSeek, OpenAI, Qwen, Doubao)
- v2.1 — Cost-per-task attribution with ROI tracking
- v2.2 — Team/multi-agent fleet monitoring dashboard
- v3.0 — Predictive spend forecasting with anomaly detection
We welcome contributions. If you're running OpenClaw agents in production and want to help make Cost Guard better, open an issue or submit a PR.
Priority areas:
- Multi-provider API adapters (DeepSeek, OpenAI, etc.)
- Additional messaging channel integrations
- Internationalization (language packs beyond EN/中文)
Cost Guard 是一个用于 OpenClaw 代理的自主成本监控技能。它可以实时跟踪 AI 代理的 API 支出，在成本失控之前捕获问题，并自动生成每日支出报告。由 Lemnos AI 构建 — 运营优先的 AI 公司。
安装:
openclaw skill install cost-guard详细中文文档请参阅 README_ZH.md
MIT License. Use it, fork it, build on it. Just don't let your agents burn cash. 🦞
Built with 🦞 by Lemnos AI
253 operators trust Cost Guard to protect their AI agents.