📄 中文摘要
大多数团队无需等待SDK封装即可获得清晰的成本可视化。通过直接的摄取合同和安全的异步发送器，可以立即实现有用的LLM成本峰值检测。该设置提供了端点级成本归属、租户/用户集中视图、提示部署回归检测以及预算和支出警报工作流，而无需改变提供商的流量路径。无SDK的概念并不意味着永远手动操作，而是保持提供商调用不变，从提供商响应中提取使用元数据，并异步发送标准化的遥测负载。未来的SDK封装可以减少重复工作。
无SDK LLM 生产环境中的成本峰值检测（端点 + 用户 + 提示版本）
📄 English Summary
No-SDK LLM Cost Spike Detection in Production (Endpoint + User + PromptVersion)
Most teams can achieve significant cost visibility without waiting for SDK wrappers. A practical setup allows for effective LLM cost spike detection using a direct ingest contract and a safe async sender. This setup provides endpoint-level cost attribution, tenant/user concentration views, prompt deploy regression detection, and budget and spend-alert workflows without altering provider traffic paths. The term 'No-SDK' does not imply manual processes indefinitely; rather, it involves keeping provider calls unchanged, extracting usage metadata from provider responses, and sending a normalized telemetry payload asynchronously. Future SDK wrappers can streamline this process further.