agentpulse
由 sru4ka 提供
AI & LLMs简介
Track LLM API costs, tokens, latency, and errors for your AI agent.
描述
该技能的功能
AgentPulse tracks LLM API calls made by your agent, recording costs, token counts, latency, and errors. Data is sent to a real-time dashboard at agentpulse.dev. Cost is calculated server-side against an up-to-date pricing table covering 50+ models across Anthropic, OpenAI, Google, MiniMax, DeepSeek, and Mistral.
Cost is calculated server-side using a maintained pricing table, so you don't need to track per-model pricing yourself or write custom logging code.
何时使用
- Checking daily or weekly API spend across models
- Identifying which model is driving the most cost
- Reviewing error rates and failed API calls
- Monitoring average response latency over time
- Setting alerts when daily spend exceeds a budget
示例工作流
User asks: How much have I spent this week?
- 1
Call GET https://agentpulse.dev/api/stats/overview with the API key
- 2
Parse the JSON response for daily_stats and total_cost fields
- 3
Sum costs from daily_stats entries covering the past 7 days
- 4
Present the weekly total and a per-day breakdown to the user
This week you spent $4.32 across 847 API calls. Monday was your heaviest day at $1.10, mostly from claude-sonnet-4-5.
前置要求
- AgentPulse account (free signup at agentpulse.dev)
- AGENTPULSE_API_KEY environment variable
安装
clawhub install sru4ka/agentpulse运行 CLI 命令,或把这段提示词粘贴到 Claude Code、Cursor、Codex 或你在用的任意编程智能体里。