跳到技能内容
词映 AI导航
发现工具选型指南Skills 技能AI 快讯AI图片AI视频AI文档AI办公AI助手AI搜索AI编程AI音频AI设计AI智能体模型与开发AI学习模型评测AI提示词内容检测AI金融AI 3D扫描与识别其他工具跨境电商 AI
词映 AI导航
工具库Skills 技能
我的收藏
词映 AI导航
发现工具选型指南Skills 技能AI 快讯AI图片AI视频AI文档AI办公AI助手AI搜索AI编程AI音频AI设计AI智能体模型与开发AI学习模型评测AI提示词内容检测AI金融AI 3D扫描与识别其他工具跨境电商 AI
Skills directory/sampleXbro/agentsmesh/am-agent-code-debugger

Codex AgentsMesh Debugging for State and Root-Cause Evidence

Reduce a local bug to repeatable input and test hypotheses using logs, breakpoints, and state observations.

中文

Install in Codex

Codex with an AgentsMesh or target checkout, a short repro, relevant logs, and non-sensitive test data.

  1. Open the pinned GitHub original and retain the complete `.agents/skills/am-agent-code-debugger/` directory in the checkout.
  2. Open Codex in that checkout and invoke `$am-agent-code-debugger` with the example task and desired output.

Example task

Give `$am-agent-code-debugger` a fictional local list-sort issue. Compare empty input with two records and identify the first state divergence.

What to check

The report records reproduction, evidence, cause or remaining hypotheses, and a repeatable regression check.

Original SKILL.md on GitHub · Updated 2026-11-13

Frequently asked question

Which error matters when several logs fail?

Trace the earliest unexpected state for the same input. Later logs can show how it propagated.

Related skills from this repository

  • Professional Engineering Communication Skill for ChatGPT: Write Clear Commit Messages
Browse Codex skills in the Chinese directory
词映 AI 导航Skills 技能
工具分类全部工具已本地化技能选型指南AI 快讯