rajudandigam

agent-inspect

rajudandigam

Local execution trees for TypeScript AI agents. agent-inspect helps you understand what happened inside an AI agent run — locally. It turns manual steps, tool calls, LLM calls, structured logs, failures, durations, and run metadata into readable execution trees you can inspect from the terminal. It is built for TypeScript/Node.js developers..

AI 简介

agent-inspect 是一个用于 TypeScript AI 代理的本地执行树工具。它能够将手动步骤、工具调用、大语言模型调用、结构化日志、失败信息、持续时间和运行元数据转换为可从终端查看的易读执行树,帮助开发者理解 AI 代理在本地运行时的行为。该工具专为 TypeScript/Node.js 开发者设计,适用于需要在本地调试或快速迭代 AI 代理产品的场景,且无需云端上传或第三方服务接入。此外,agent-inspect 支持与现有结构化日志集成,并提供标准对齐的本地导出选项,确保了开发流程中的灵活性和独立性。

TypeScript
MIT License
95
Stars
81
Forks
72
Watchers
17
Issues

Star 增长

今日0
近 7 天+3
近 30 天+3
综合评分47.54
默认分支main