Token导航 LogoToken导航TokenDH.com
研究检索只读github未标认证来源可访问许可证需确认审计提醒

debug-tools调试工具

Agent Skill

debug-tools 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

1,430

周安装

59

GitHub Stars

5

下载量

467
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:debug-tools(调试工具)
来源仓库:https://github.com/adeonir/agent-skills
仓库路径:skills/debug-tools
安装命令:
npx skills add https://github.com/adeonir/agent-skills --skill debug-tools
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/adeonir/agent-skills --skill debug-tools

简介

灵活的迭代式调试工作流,支持多种技术组合直至问题解决或升级处理。

  • 适用于多轮修复尝试未果后的策略调整与知识沉淀需求。
  • 自动清理冗余日志并在验证成功后结束循环进入交付阶段。
  • 最多允许三轮尝试,超出阈值则记录结论并转交人工深度分析。
  • debug-tools 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Debug Tools

Iterative debugging workflow with flexible technique selection and escalation.

Workflow

investigate --> fix --> verify --> done
  ^_______________________|  (max 3 attempts, then escalate)

Core loop: investigate, fix, verify. Techniques (log injection, pattern comparison, focus area analysis) are tools within investigation, not mandatory phases. Log cleanup happens automatically after verification succeeds.

Context Loading Strategy

Load only the reference matching the current trigger. Multiple references may be loaded during a full debugging session (investigation often leads to log injection).

Triggers

Trigger PatternReference
Debug issue, investigate bug, fix buginvestigation.md/debug-tools/references/investigation.md)
Add debug logs, inject logs, trace with logslog-injection.md/debug-tools/references/log-injection.md)
Remove debug logs, cleanup logslog-cleanup.md/debug-tools/references/log-cleanup.md)
Debug patterns, log format, common bugsdebugging-patterns.md/debug-tools/references/debugging-patterns.md)

Cross-References

investigation.md <---> log-injection.md (investigation may request logs)
investigation.md <---> log-cleanup.md (after fix verified)
investigation.md <---> debugging-patterns.md (pattern comparison)
log-injection.md ----> log-cleanup.md (cleanup removes injected logs)

Guidelines

DO:

  • Use confidence scoring: >= 70 report as probable cause, 50-69 suggest logs
  • Compare broken code against working examples when root cause is unclear
  • Always use [DEBUG] prefix for injected logs (enables cleanup)
  • Apply minimal fix: smallest change that resolves the issue
  • Redact sensitive values (passwords, tokens, PII) in injected logs
  • Clean up debug logs automatically after fix is verified
  • Track fix attempts: after 3 failed fixes, escalate on the 4th to architectural review
  • Use whatever debugging tools are available in the environment

DON'T:

  • Report findings with confidence < 50 (contrasts: >= 70 probable cause, 50-69 suggest logs)
  • Log sensitive data (contrasts: redact sensitive values in injected logs)
  • Apply large refactors as part of a bug fix (contrasts: smallest change that resolves the issue)
  • Leave debug logs in production code (contrasts: clean up debug logs after verification)
  • Keep retrying the same approach when fixes fail repeatedly (contrasts: track attempts and escalate on the 4th)

Error Handling

  • No bug description provided: ask user to describe the issue
  • Cannot reproduce: suggest adding debug logs
  • Fix doesn't work: return to investigation with new evidence
  • Three failed fix attempts: escalate to architectural review on the next attempt
  • Logs left behind: user can request cleanup anytime

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

38.27%
按下载量换算179

Claude

30.18%
按下载量换算141

Cursor

18.43%
按下载量换算86

Gemini CLI

10.7%
按下载量换算50

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills