🧪 Skills
Model Usage Monitor
监控并统计模型调用次数和成本,计算缓存命中率,支持实时监控与每小时自动告警。
v1.0.1
Description
model-usage-monitor Skill
OpenClaw 模型使用监控与告警技能
⚠️ Security & Permissions Declaration
This skill monitors log files and sends alerts. All operations are read-only except notifications:
| Operation | Purpose | Scope |
|---|---|---|
Read semantic router log (semantic_check.log) |
Parse model usage statistics | Read-only, local file |
| Read OpenClaw logs | Detect usage anomalies | Read-only, local files |
| Send alert notifications via OpenClaw messaging | Notify user of cost spikes | Local OpenClaw API only |
What this skill does NOT do:
- Does NOT modify any configuration or log files
- Does NOT access external servers or APIs
- Does NOT access model credentials directly
- Does NOT require elevated privileges
- Read-only monitoring — zero side effects on system state
功能
- 解析语义路由日志,统计模型使用分布
- 估算各模型调用次数和成本
- 计算缓存命中率
- 每小时自动告警检查
- 支持实时监控模式
安装
# 技能已包含监控脚本和自动配置
# 安装后自动创建每小时检查的 Cron Job
使用
查看监控报告
# 完整报告
python3 ~/.openclaw/workspace/.lib/model_usage_monitor_v2.py
# JSON 格式
python3 ~/.openclaw/workspace/.lib/model_usage_monitor_v2.py --format json
# 仅检查告警
python3 ~/.openclaw/workspace/.lib/model_usage_monitor_v2.py --alert-check
实时监控
python3 ~/.openclaw/workspace/.lib/model_usage_monitor_v2.py --live
告警阈值
| 类型 | 阈值 | 说明 |
|---|---|---|
| Opus 调用频繁 | >5 次/小时 | 防止意外大量使用昂贵模型 |
| Opus 成本过高 | >$0.50/小时 | 成本控制 |
| 总成本过高 | >$2.00/小时 | 总体预算控制 |
文件结构
.skills/model-usage-monitor/
├── SKILL.md # 本文件
├── monitor.py # 核心监控脚本
├── setup.py # 自动安装/配置
└── config.json # 默认配置
技术细节
- 全部使用本地 Ollama 模型,零 API 调用
- 只读日志文件,零侵入
- 基于 semantic_check.log 和 gateway.log 分析
Reviews (0)
Sign in to write a review.
No reviews yet. Be the first to review!
Comments (0)
No comments yet. Be the first to share your thoughts!