Debug System
Find, explain, and fix errors by identifying the root cause.
Use this skill when:
- - something is not working
- an error message appears
- a workflow fails
- an API request breaks
- behavior does not match expectations
- the same issue keeps happening
This is an instruction-only skill.
It does not execute code or access external systems unless the user provides the data.
What this skill does
This skill helps an agent:
- - analyze errors and unexpected behavior
- identify the real root cause (not just symptoms)
- explain what went wrong in simple terms
- suggest concrete fixes
- guide step-by-step debugging
- prevent the issue from happening again
What this skill does not do
This skill does not:
- - guess without evidence
- invent missing logs or results
- claim something is fixed without validation
- access external systems without user input
- provide fake certainty
Inputs
Use what the user provides:
- - error messages
- logs
- code snippets
- API responses
- workflow steps
- expected vs actual behavior
- screenshots or descriptions
If information is missing, state assumptions clearly.
Debugging process
Step 1 — Understand the problem
Identify:
- - expected behavior
- actual behavior
- when the issue occurs
- what changed recently
Step 2 — Analyze the error
Look for:
- - exact error message
- stack trace (if available)
- failing step or node
- invalid inputs
- missing fields
- wrong data types
- permission/auth issues
- network/API failures
Step 3 — Find root cause
Do not stop at symptoms.
Ask:
- - what caused this error?
- what condition triggered it?
- is this reproducible?
- is it a config issue, logic issue, or external failure?
Step 4 — Propose fixes
For each issue, provide:
- - root cause
- why it happens
- exact fix
- alternative fix (if needed)
- how to verify the fix
Step 5 — Prevent recurrence
Suggest:
- - validation checks
- better error handling
- logging improvements
- retry mechanisms
- safer defaults
Output format
Debug Report
Problem Summary
- - Expected:
- Actual:
- Context:
Root Cause
Evidence
Fix
Recommended Fix
Alternative Fix
Verification Steps
- 1. ...
- ...
- ...
Prevention
Open Questions
Behavior rules
- - Focus on root cause, not just symptoms
- Be clear and actionable
- Do not assume missing data without stating it
- Prefer simple fixes over complex ones
- Explain in a way the user can follow step-by-step
Quality checklist
Before finishing, ensure:
- - root cause is identified
- fix is actionable
- verification steps are clear
- no fake assumptions
- prevention is included
When not to use this skill
Do not use this skill for:
- - general explanations without a problem
- theoretical discussions without real issues
- tasks unrelated to debugging
Use only when there is a failure, bug, or unexpected behavior.
调试系统
通过识别根本原因来查找、解释和修复错误。
在以下情况下使用此技能:
- - 某些功能无法正常工作
- 出现错误消息
- 工作流执行失败
- API请求中断
- 行为与预期不符
- 同一问题反复出现
此技能仅为指令型技能。
除非用户提供数据,否则不会执行代码或访问外部系统。
此技能的功能
此技能帮助智能体:
- - 分析错误和意外行为
- 识别真正的根本原因(而不仅仅是表面症状)
- 用通俗语言解释问题所在
- 提出具体的修复建议
- 指导逐步调试过程
- 防止问题再次发生
此技能不做什么
此技能不会:
- - 在没有证据的情况下进行猜测
- 虚构缺失的日志或结果
- 未经验证就声称问题已修复
- 在没有用户输入的情况下访问外部系统
- 提供虚假的确定性
输入内容
使用用户提供的信息:
- - 错误消息
- 日志
- 代码片段
- API响应
- 工作流步骤
- 预期行为与实际行为
- 截图或描述
如果信息缺失,需明确说明假设条件。
调试流程
第一步 — 理解问题
识别:
第二步 — 分析错误
查找:
- - 确切的错误消息
- 堆栈跟踪(如有)
- 失败的步骤或节点
- 无效输入
- 缺失字段
- 错误的数据类型
- 权限/认证问题
- 网络/API故障
第三步 — 查找根本原因
不要停留在表面症状上。
提问:
- - 是什么导致了此错误?
- 什么条件触发了它?
- 这个问题是否可以重现?
- 是配置问题、逻辑问题还是外部故障?
第四步 — 提出修复方案
针对每个问题,提供:
- - 根本原因
- 发生原因
- 确切修复方法
- 替代修复方案(如有需要)
- 如何验证修复效果
第五步 — 防止复发
建议:
- - 验证检查
- 更好的错误处理
- 日志改进
- 重试机制
- 更安全的默认值
输出格式
调试报告
问题摘要
根本原因
证据
修复方案
推荐修复
替代修复
验证步骤
- 1. ...
- ...
- ...
预防措施
待解决问题
行为规则
- - 关注根本原因,而非表面症状
- 清晰且可操作
- 不要在没有说明的情况下假设缺失数据
- 优先选择简单修复而非复杂方案
- 以用户能逐步理解的方式解释
质量检查清单
完成前确保:
- - 已识别根本原因
- 修复方案可操作
- 验证步骤清晰
- 无虚假假设
- 包含预防措施
何时不使用此技能
以下情况不要使用此技能:
- - 没有具体问题的一般性解释
- 没有实际问题的理论讨论
- 与调试无关的任务
仅在出现故障、错误或意外行为时使用。