Featured

Deploy OpenClaw in 60 seconds — 20% off logoDeploy OpenClaw in 60 seconds — 20% off

Launch OpenClaw on Hostinger in about 60 seconds and keep your agent live 24/7. Our referral link gives you 20% off, no coupon code needed.

Launch on Hostinger
Run your Hermes agent on Hostinger, fully managed logoRun your Hermes agent on Hostinger, fully managed

Launch Hermes on Hostinger in one click, fully managed, no VPS knowledge needed. Use code ZACAARON10 for 10% off.

Launch on Hostinger
Crawl and scrape any site into clean data, 10% off logoCrawl and scrape any site into clean data, 10% off

Firecrawl crawls and scrapes any site into clean markdown for your agent. Get 1,000 free credits, and new users get 10% off their first purchase.

Try Firecrawl free
6,000+ web scrapers for your AI agent, start free logo6,000+ web scrapers for your AI agent, start free

Apify gives your agent live web data: 6,000+ prebuilt scrapers and actors, MCP-ready. Sign up free with $5 in usage credits.

Try Apify free
One API to scrape, enrich, and extract the internet. logoOne API to scrape, enrich, and extract the internet.

Context.dev gives your agents a single API to scrape, enrich, and extract live web data — no proxies, no parsers, no maintenance.

Start building free
SetupClaw: done-for-you OpenClaw for founders & exec teams logoSetupClaw: done-for-you OpenClaw for founders & exec teams

White-glove OpenClaw for founders and exec teams (4–50+ employees): we install, harden, integrate your tools, and maintain it — secured from day one.

Get it set up for you
SEO data APIs for your agent, $1 free credit logoSEO data APIs for your agent, $1 free credit

DataForSEO gives your agent live access to SERP results, keyword data, backlinks, and on-page SEO data through one API. New accounts get a $1 credit, good for up to 20,000 keyword or backlink lookups.

Try DataForSEO free
Reach 48,000+ AI builders

A flat monthly placement in front of developers actively installing AI tools. No lock-in, cancel anytime.

Advertise here

Works with

Claude CodeClaude DesktopCursorVS CodeClineCodex CLIOpenClaw+ any MCP client

Install to Claude Code

This server doesn't publish a one-line install command. Follow the setup in the source repository.

Summary

A smart debugging agent that remembers errors, automatically diagnoses issues, and accumulates solutions for faster resolution.

README.md

Debug MCP - 智能调试 Agent

一个会"记住错误"的智能调试工具,自动排查问题并积累解决方案。

特点

  • 🔍 自动排查 - 智能分析错误,定位根因
  • 📚 错误记忆 - 自动保存排查记录,下次类似问题秒解
  • 🛡️ 主动预防 - 代码预检,提前发现风险
  • 📊 趋势分析 - 了解错误模式针对性学习
  • 质量评分 - 高评价方案优先推荐
  • 🧠 ReAct 推理 - 思考 → 行动 → 观察 → 反思
  • 🔌 MCP 协议 - 支持 Claude Desktop、Cursor
  • 🌐 多 LLM - DeepSeek / OpenAI / Anthropic
  • 📁 无需数据库 - 纯 JSON 文件存储案例

---

新人使用步骤

1. 安装

git clone https://github.com/你的用户名/debug-mcp.git
cd debug-mcp
pip install -e .

2. 配置 API Key(二选一)

方式一:创建 .env 文件 ```bash cp .env.example .env

编辑 .env,填入你的 DEEPSEEK_API_KEY


**方式二:直接传入**

agent = DebugAgent(api_key="sk-your-key") ```

3. 使用(两种方式)

方式 A:Python 直接调用(推荐)

from src.agent import DebugAgent

agent = DebugAgent()

# 排查问题
result = agent.debug("TypeError: Cannot read property 'id' of undefined")

print(result)

方式 B:MCP Server(需要 Claude Desktop)

配置 claude_desktop_config.json

{
  "mcpServers": {
    "debug-mcp": {
      "command": "python",
      "args": ["-m", "src.server"]
    }
  }
}

重启 Claude Desktop,然后直接说:

  • "排查一下这个错误"
  • "看看这个 bug"

---

MCP 工具列表

| 工具 | 说明 | |------|------| | debug | 排查问题 - 输入错误信息,返回解决方案 | | search_case | 搜索历史案例 | | list_cases | 列出所有案例 | | get_case | 查看案例详情 | | delete_case | 删除案例 | | mark_effective | 标记方案有效性(帮助改进匹配) | | get_recommended_fixes | 获取高评价解决方案 | | pre_check_code | 代码风险预检(主动预防) | | get_weekly_report | 获取本周错误报告 | | get_error_trends | 获取错误趋势分析 | | get_stats | 统计信息 | | clear_memory | 清空记忆 | | search_code | 搜索代码文件 | | read_file | 读取文件内容 | | grep | 正则搜索 | | check_syntax | 语法检查 | | list_files | 列出文件 | | refresh_index | 刷新索引 |

---

如何避免重复犯错?

使用以下 5 个最佳实践:

1️⃣ 描述错误要具体

# ❌ 太笼统
agent.debug("程序出错了")

# ✅ 具体描述
agent.debug("TypeError: Cannot read property 'id' of undefined")

2️⃣ 看到 found_in_history: True 直接用历史方案

result = agent.debug("Cannot read property 'id' of undefined")

# 如果 found_in_history: True
# 直接使用 result['solution'],无需重新排查

3️⃣ 定期查看高频错误

# 查看最常遇到的错误,针对性预防
agent.list_cases(limit=10)  # 高频错误排行
agent.get_stats()           # 统计信息
agent.get_weekly_report()   # 本周报告

4️⃣ 使用预检主动预防

# 在编码时主动检查风险
agent.pre_check(code="your_code_here")

# 或使用 MCP
# "检查一下这段代码有没有风险"

5️⃣ 标记方案有效性帮助改进

# 如果方案有效
agent.memory.mark_effective(case_id, effective=True)

# 如果方案无效
agent.memory.mark_effective(case_id, effective=False)

# 获取高评价方案
agent.memory.get_effective_cases(min_rating=0.5)

---

核心思想

这个 MCP 的价值在于积累

  • 用得越多,案例库越丰富
  • 标记有效性 → 匹配算法越精准
  • 定期查看错误趋势 → 针对性学习预防

---

示例

from src.agent import DebugAgent

agent = DebugAgent(api_key="sk-xxx")

# 第一次排查
result = agent.debug("TypeError: Cannot read property 'id' of undefined")
# 输出:
# {
#   "success": True,
#   "root_cause": "接口返回数据为null时未做空值检查",
#   "fix_solution": "使用 data?.id 或 data || {}",
#   "steps": [{"action": "...", "observation": "..."}],
#   "found_in_history": False
# }

# 第二次排查相同错误(自动匹配历史)
result = agent.debug("Cannot read property 'id' of undefined")
# 输出:
# {
#   "success": True,
#   "found_in_history": True,
#   "fix_solution": "使用 data?.id 或 data || {}",
#   "history_case": {...}
# }

---

项目结构

debug-mcp/
├── src/
│   ├── agent.py        # Debug Agent 核心
│   ├── memory.py       # 案例库(JSON 文件)
│   ├── tools.py        # 工具集
│   └── server.py       # MCP Server
├── cases/              # 案例存储目录(自动创建)
│   └── debug_cases.json
└── .env               # API Key 配置

---

案例库

  • 位置:cases/debug_cases.json
  • 无需数据库,纯文件存储
  • 每次排查自动保存
  • 下次遇到类似问题自动匹配

---

API

from src.agent import DebugAgent

agent = DebugAgent(api_key="sk-xxx")

# 排查问题
result = agent.debug("错误信息")

# 搜索历史案例
cases = agent.search_history(["关键词"])

# 获取统计
stats = agent.get_stats()

# 主动预防:检查代码风险
result = agent.pre_check(code="your code here")

# 获取高评价方案
effective_cases = agent.memory.get_effective_cases(min_rating=0.5)

# 标记方案是否有效
agent.memory.mark_effective(case_id, effective=True)

# 获取周报
weekly_report = agent.memory.get_weekly_report()

# 获取趋势分析
trends = agent.memory.get_error_trends(days=30)

# 清空记忆
agent.clear_memory()

---

配置选项

agent = DebugAgent(
    api_key="sk-xxx",           # API Key(必须)
    model="deepseek-chat",      # 模型,默认 deepseek-chat
    max_steps=5,                # 最大排查步骤
    case_file="cases/debug_cases.json"  # 案例库路径
)

---

支持的模型

| 模型 | 配置 | |------|------| | DeepSeek(默认) | model="deepseek-chat" | | OpenAI | model="gpt-4" | | Anthropic | model="claude-3-opus" | | Ollama | model="llama2" |

---

给 Claude 的系统规则

如果你是用户,可以在对话中告诉 Claude 以下规则(让它帮你解决问题时更聪明):

你是一个调试助手。在解决问题时:
1. 每次尝试新方法前,先问用户确认
2. 如果一个方法失败,不要用相同方法重试
3. 可以调用 debug-mcp 预检工具检查风险
4. 避免重复尝试已经失败的方法
5. 遇到不确定的问题,先搜索历史案例

让 Claude 每次尝试前先用 pre_check_code 检查一下代码风险。

---

有问题?直接在项目中提 Issue!

See related servers & alternatives →

Related MCP servers

Browse all →

Related guides

Hand-picked reading to help you choose and use Developer Tools servers.