Zen MCP Server E6de5294

MCP开发工具

提供方 passionmuse16

开发工具 未标注 API Key 社区 / 第三方 Claude DesktopClaude

Zen MCP是一个AI模型协调服务器,提供多模型协作开发环境,支持代码分析、问题解决和协作开发,适用于复杂软件开发流程。

一句话选型结论

按资料字段整理

Zen MCP Server E6de5294 属于开发工具类 MCP 服务器,传输方式未标注,API Key,可直接接入 Claude Desktop、Claude,没有现成安装命令,需按官方文档接入。接入前先确认权限范围;涉及写入、删除或付款的操作,建议先在测试环境验证。

分类开发工具 传输方式未标注 鉴权API Key 部署方式未标注
GitHub 星标 2 仓库热度,无公开仓库时显示未标注
工具数 16 暴露给模型的工具数量
资源 / 提示词 0 / 0 资源与提示词模板数量
维护方 社区 / 第三方 核验等级 仅资料核对

工具与能力

开发工具
数量来自来源目录的标注,具体工具清单以官方文档为准
工具数量16暴露给模型可调用的工具个数
资源数量0可读取的资源条目
提示词数量0内置提示词模板数量

能力标签

支持客户端

适用场景

AI协作Python本地部署代码分析多模型协调开发工具自动化编程

安装与配置

未标注 API Key
上游未给出传输方式
  1. 1
    准备运行环境

    按官方文档确认运行环境(常见为 Node.js 或 Python),装好后再执行安装命令。

  2. 2
    安装服务

    复制上面的安装命令执行;命令里带占位符的位置,换成自己的密钥或路径。

  3. 3
    写入客户端配置

    在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,让新配置生效。

  4. 4
    验证是否可用

    连接成功后客户端会列出约 16 个工具,随便调一个只读工具确认能返回结果。

典型使用场景

3 类
按分类与标签整理的常见用法,具体以官方文档为准
代码与仓库操作

在对话里直接查代码、提 issue、走 PR,省掉来回切工具。

本地命令与脚本

把构建、测试、格式化这类重复命令交给模型执行。

接口联调与测试

让模型调你的接口、跑一遍用例,快速定位问题。

选择建议

选之前先看这几条
按条目字段给出的判断,不替你决定用不用

值得选的地方

  • 资料标注 16 个工具,覆盖面较广,能满足多种调用。
  • 已适配 Claude Desktop、Claude,换客户端不用重新研究配置。

先想清楚的地方

  • 没有现成安装命令,需要按官方文档自己填启动命令和参数。
  • 使用前要准备好API Key对应的凭据,并按最小权限配置。
  • 版本更新与行为细节以官方仓库为准。

更适合

  • 需要在开发工具场景里让模型直接调用外部能力的人。
  • 已经在用 Claude Desktop、Claude 的用户,接入路径最短。
  • 想先快速试一下、能接受在本地或自有环境跑服务的人。

不太适合

  • 不方便管理密钥或第三方授权的人。
  • 需要厂商级 SLA 和技术支持的生产环境。

同类对比

6 个同类
同分类里挑最接近的几条,字段以各自详情页为准
MCP 服务器工具数传输方式部署方式安装命令最后核验
Zen MCP Server E6de5294当前 16 未标注 未标注 无 2026-05-17
Git 838e2ec1 12 stdio · 本地进程 本地部署 有 2026-05-17
Markitdown MCP 2c653a5f 1 stdio · 本地进程 本地部署 有 2026-05-17
JavaGuide 未标注 stdio · 本地进程 本地部署 有 2026-04-11
awesome-claude-skills 未标注 stdio · 本地进程 本地部署 有 2026-04-26
Context7 831acf40 未标注 stdio · 本地进程 本地部署 有 2026-05-17

服务器档案

开发工具
字段来自项目仓库与公开索引,中文说明由本站标注
提供方passionmuse16上游标注的维护方 passionmuse16
分类开发工具本站按上游分类展示
连接方式未标注上游未给出传输方式
鉴权方式API Key远程服务建议用最小权限凭据
维护方社区 / 第三方核验等级 仅资料核对
包名 / 运行时未标注 · 未标注本地安装时用到的包标识与运行时
部署类型未标注上游给出的部署形态
最近核验2026-05-17上游最后核验时间

详细介绍

35.8 KB
来自项目仓库与上游目录的原文,未做删改

Zen MCP: Many Workflows. One Context.

zen_web.webm

🤖 Claude OR Gemini CLI + [Gemini / OpenAI / Grok / OpenRouter / DIAL / Ollama / Any Model] = Your Ultimate AI Development Team

The ultimate development partners for your favorite Coding Agent (Claude OR Gemini CLI) - a Model Context Protocol server that gives you access to multiple AI
models for enhanced code analysis, problem-solving, and collaborative development.

Features true AI orchestration with conversations that continue across workflows - Give Claude a complex
workflow and let it orchestrate between models automatically. Claude stays in control, performs the actual work,
but gets perspectives from the best AI for each subtask. With tools like planner for
breaking down complex projects, analyze for understanding codebases,
codereview for audits, refactor for
improving code structure, debug for solving complex problems, and precommit for
validating changes, Claude can switch between different tools and models mid-conversation,
with context carrying forward seamlessly.

Example Workflow - Claude Code:
1. Perform a codereview using gemini pro and o3 and use planner to generate a detailed plan, implement the fixes and do a final precommit check by continuing from the previous codereview
2. This triggers a codereview workflow where Claude walks the code, looking for all kinds of issues
3. After multiple passes, collects relevant code and makes note of issues along the way
4. Maintains a confidence level between exploring, low, medium, high and certain to track how confidently it's been able to find and identify issues
5. Generates a detailed list of critical -> low issues
6. Shares the relevant files, findings, etc with Gemini Pro to perform a deep dive for a second codereview
7. Comes back with a response and next does the same with o3, adding to the prompt if a new discovery comes to light
8. When done, Claude takes in all the feedback and combines a single list of all critical -> low issues, including good patterns in your code. The final list includes new findings or revisions in case Claude misunderstood or missed something crucial and one of the other models pointed this out
9. It then uses the planner workflow to break the work down into simpler steps if a major refactor is required
10. Claude then performs the actual work of fixing highlighted issues
11. When done, Claude returns to Gemini Pro for a precommit review

All within a single conversation thread! Gemini Pro in step 11 knows what was recommended by O3 in step 7! Taking that context
and review into consideration to aid with its final pre-commit review.

Think of it as Claude Code for Claude Code. This MCP isn't magic. It's just super-glue.

> Remember: Claude stays in full control — but YOU call the shots.
> Zen is designed to have Claude engage other models only when needed — and to follow through with meaningful back-and-forth.
> You're the one who crafts the powerful prompt that makes Claude bring in Gemini, Flash, O3 — or fly solo.
> You're the guide. The prompter. The puppeteer.
> ### You are the AI - Actually Intelligent.

Because these AI models clearly aren't when they get chatty →

Quick Navigation

  • Getting Started
  • Quickstart - Get running in 5 minutes
  • Available Tools - Overview of all tools
  • AI-to-AI Conversations - Multi-turn conversations

  • Tools Reference

  • chat - Collaborative thinking
  • thinkdeep - Extended reasoning
  • challenge - Prevents You're absolutely right! responses
  • planner - Interactive step-by-step planning
  • consensus - Multi-model consensus analysis
  • codereview - Code review
  • precommit - Pre-commit validation
  • debug - Debugging help
  • analyze - File analysis
  • refactor - Code refactoring with decomposition focus
  • tracer - Call-flow mapping and dependency tracing
  • testgen - Test generation with edge cases
  • secaudit - Security audit with OWASP analysis
  • docgen - Documentation generation with complexity analysis

  • Advanced Usage

  • Advanced Features - AI-to-AI conversations, large prompts, web search
  • Complete Advanced Guide - Model configuration, thinking modes, workflows, tool parameters

  • Setup & Support

  • WSL Setup Guide - Windows Subsystem for Linux configuration
  • Troubleshooting Guide - Common issues and debugging steps
  • License - Apache 2.0

Why This Server?

Claude is brilliant, but sometimes you need:
- Multiple AI perspectives - Let Claude orchestrate between different models to get the best analysis
- Automatic model selection - Claude picks the right model for each task (or you can specify)
- A senior developer partner to validate and extend ideas (chat)
- A second opinion on complex architectural decisions - augment Claude's thinking with perspectives from Gemini Pro, O3, or dozens of other models via custom endpoints (thinkdeep)
- Get multiple expert opinions - Have different AI models debate your ideas (some supporting, some critical) to help you make better decisions (consensus)
- Professional code reviews with actionable feedback across entire repositories (codereview)
- Pre-commit validation with deep analysis using the best model for the job (precommit)
- Expert debugging - O3 for logical issues, Gemini for architectural problems (debug)
- Extended context windows beyond Claude's limits - Delegate analysis to Gemini (1M tokens) or O3 (200K tokens) for entire codebases, large datasets, or comprehensive documentation
- Model-specific strengths - Extended thinking with Gemini Pro, fast iteration with Flash, strong reasoning with O3, local privacy with Ollama
- Local model support - Run models like Llama 3.2 locally via Ollama, vLLM, or LM Studio for privacy and cost control
- Dynamic collaboration - Models can request additional context and follow-up replies from Claude mid-analysis
- Smart file handling - Automatically expands directories, manages token limits based on model capacity
- Vision support - Analyze images, diagrams, screenshots, and visual content with vision-capable models
- Bypass MCP's token limits - Work around MCP's 25K limit automatically
- Context revival across sessions - Continue conversations even after Claude's context resets, with other models maintaining full history

Pro Tip: Context Revival

This is an extremely powerful feature that cannot be highlighted enough:

> The most amazing side-effect of this conversation continuation system is that even AFTER Claude's context resets or
> compacts, since the continuation info is kept within MCP's memory, you can ask it to continue discussing
> the plan with o3, and it will suddenly revive Claude because O3 would know what was being talked about and
> relay this back in a way that re-ignites Claude's understanding. All this without wasting context on asking Claude to
> ingest lengthy documents / code again and re-prompting it to communicate with another model. Zen manages that internally. The model's response
> revives Claude with better context around the discussion than an automatic summary ever can.

📖 Read the complete technical deep-dive on how this revolutionary system works

This server orchestrates multiple AI models as your development team, with Claude automatically selecting the best model for each task or allowing you to choose specific models for different strengths.

Prompt Used:`` Study the code properly, think deeply about what this does and then see if there's any room for improvement in terms of performance optimizations, brainstorm with gemini on this to get feedback and then confirm any change by first adding a unit test withmeasure` and measuring current code and then implementing the optimization and
measuring again to ensure it improved, then share results. Check with gemini in between as you make tweaks.


The final implementation resulted in a 26% improvement in JSON parsing performance for the selected library, reducing processing time through targeted, collaborative optimizations guided by Gemini’s analysis and Claude’s refinement.

## Quickstart (5 minutes)

### Prerequisites

- Python 3.10+ (3.12 recommended)
- Git
- **Windows users**: WSL2 is required for Claude Code CLI

### 1. Get API Keys (at least one required)

**Option A: OpenRouter (Access multiple models with one API)**
- **OpenRouter**: Visit OpenRouter for access to multiple models through one API. Setup Guide
  - Control model access and spending limits directly in your OpenRouter dashboard
  - Configure model aliases in `conf/custom_models.json`

**Option B: Native APIs**
- **Gemini**: Visit Google AI Studio and generate an API key. For best results with Gemini 2.5 Pro, use a paid API key as the free tier has limited access to the latest models.
- **OpenAI**: Visit OpenAI Platform to get an API key for O3 model access.
- **X.AI**: Visit X.AI Console to get an API key for GROK model access.
- **DIAL**: Visit DIAL Platform to get an API key for accessing multiple models through their unified API. DIAL is an open-source AI orchestration platform that provides vendor-agnostic access to models from major providers, open-source community, and self-hosted deployments. API Documentation

**Option C: Custom API Endpoints (Local models like Ollama, vLLM)**
Please see the setup guide. With a custom API you can use:
- **Ollama**: Run models like Llama 3.2 locally for free inference
- **vLLM**: Self-hosted inference server for high-throughput inference
- **LM Studio**: Local model hosting with OpenAI-compatible API interface
- **Text Generation WebUI**: Popular local interface for running models
- **Any OpenAI-compatible API**: Custom endpoints for your own infrastructure

> **Note:** Using multiple provider options may create ambiguity about which provider / model to use if there is an overlap.
> If all APIs are configured, native APIs will take priority when there is a clash in model name, such as for `gemini` and `o3`.
> Configure your model aliases and give them unique names in `conf/custom_models.json`

### 2. Choose Your Installation Method

**Option A: Quick Install with uvx**

**Prerequisites**: Install uv first (required for uvx)

Claude Desktop Configuration

Add this to your `claude_desktop_config.json`:```json
{
  "mcpServers": {
    "zen": {
      "command": "sh",
      "args": [
        "-c",
        "exec $(which uvx || echo uvx) --from git+https://github.com/BeehiveInnovations/zen-mcp-server.git zen-mcp-server"
      ],
      "env": {
        "PATH": "/usr/local/bin:/usr/bin:/bin:/opt/homebrew/bin:~/.local/bin",
        "OPENAI_API_KEY": "your_api_key_here"
      }
    }
  }
}

Claude Code CLI Configuration

Create a .mcp.json file in your project root for project-scoped configuration:```json
{
"mcpServers": {
"zen": {
"command": "sh",
"args": [
"-c",
"exec $(which uvx || echo uvx) --from git+https://github.com/BeehiveInnovations/zen-mcp-server.git zen-mcp-server"
],
"env": {
"PATH": "/usr/local/bin:/usr/bin:/bin:/opt/homebrew/bin:~/.local/bin",
"OPENAI_API_KEY": "your_api_key_here"
}
}
}
}


Gemini CLI Configuration

Edit `~/.gemini/settings.json` and add:```json
{
  "mcpServers": {
    "zen": {
      "command": "sh",
      "args": [
        "-c",
        "exec $(which uvx || echo uvx) --from git+https://github.com/BeehiveInnovations/zen-mcp-server.git zen-mcp-server"
      ],
      "env": {
        "PATH": "/usr/local/bin:/usr/bin:/bin:/opt/homebrew/bin:~/.local/bin",
        "OPENAI_API_KEY": "your_api_key_here"
      }
    }
  }
}

Note: While Zen MCP Server connects successfully to Gemini CLI, tool invocation is not working correctly yet. See Gemini CLI Setup for updates.

What this does:
- Zero setup required - uvx handles everything automatically
- Always up-to-date - Pulls latest version on each run
- No local dependencies - Works without Python environment setup
- Instant availability - Ready to use immediately

Option B: Traditional Clone and Set Up```bash

Clone to your preferred location

git clone https://github.com/BeehiveInnovations/zen-mcp-server.git
cd zen-mcp-server

One-command setup installs Zen in Claude

./run-server.sh

Or for Windows users using PowerShell:

./run-server.ps1

To view MCP configuration for Claude

./run-server.sh -c

PowerShell:

./run-server.ps1 -Config

See help for more

./run-server.sh --help

PowerShell:

./run-server.ps1 -Help


**What this does:**
- **Sets up everything automatically** - Python environment, dependencies, configuration
- **Configures Claude integrations** - Adds to Claude Code CLI and guides Desktop setup
- **Ready to use immediately** - No manual configuration needed
- **Also works with Gemini CLI** - See Gemini CLI Setup for configuration

**After updates:** Always run `./run-server.sh` again after `git pull` to ensure everything stays current.

**Windows users:** Using WSL? See the WSL Setup Guide for detailed instructions.

### 3. Add Your API Keys```bash
# Edit .env to add your API keys (if not already set in environment)
nano .env

# The file will contain, at least one should be set:
# GEMINI_API_KEY=your-gemini-api-key-here  # For Gemini models
# OPENAI_API_KEY=your-openai-api-key-here  # For O3 model
# OPENROUTER_API_KEY=your-openrouter-key  # For OpenRouter (see docs/custom_models.md)
# DIAL_API_KEY=your-dial-api-key-here      # For DIAL platform

# For DIAL (optional configuration):
# DIAL_API_HOST=https://core.dialx.ai      # Default DIAL host (optional)
# DIAL_API_VERSION=2024-12-01-preview      # API version (optional)
# DIAL_ALLOWED_MODELS=o3,gemini-2.5-pro    # Restrict to specific models (optional)

# For local models (Ollama, vLLM, etc.):
# CUSTOM_API_URL=http://localhost:11434/v1  # Ollama example
# CUSTOM_API_KEY=                                      # Empty for Ollama
# CUSTOM_MODEL_NAME=llama3.2                          # Default model

# Note: At least one API key OR custom URL is required

No restart needed: The server reads the .env file each time Claude calls a tool, so changes take effect immediately.

Next: Now run claude from your project folder using the terminal for it to connect to the newly added mcp server.
If you were already running a claude code session, please exit and start a new session.

If Setting up for Claude Desktop

Need the exact configuration? Run ./run-server.sh -c to display the platform-specific setup instructions with correct paths.

  1. Open Claude Desktop config: Settings → Developer → Edit Config
  2. Copy the configuration shown by ./run-server.sh -c into your claude_desktop_config.json
  3. Restart Claude Desktop for changes to take effect

4. Start Using It!

Just ask Claude naturally:
- "Think deeper about this architecture design with zen" → Claude picks best model + thinkdeep
- "Using zen perform a code review of this code for security issues" → Claude might pick Gemini Pro + codereview
- "Use zen and debug why this test is failing, the bug might be in my_class.swift" → Claude might pick O3 + debug
- "With zen, analyze these files to understand the data flow" → Claude picks appropriate model + analyze
- "Use flash to suggest how to format this code based on the specs mentioned in policy.md" → Uses Gemini Flash specifically
- "Think deeply about this and get o3 to debug this logic error I found in the checkOrders() function" → Uses O3 specifically
- "Brainstorm scaling strategies with pro. Study the code, pick your preferred strategy and debate with pro to settle on two best approaches" → Uses Gemini Pro specifically
- "First use local-llama for a quick local analysis, then use opus for a thorough security review" → Uses both providers in sequence

Available Tools

These aren't just tools—they're how you get Claude to think like a real developer. Instead of rushing to reply with
surface-level takes or shallow-insight, these workflows make Claude pause, dig into your code, and reason through
problems step by step.

It's the difference between a rushed guess and a focused second pair of eyes that actually understands your code. Try them
and feel the difference.

Quick Tool Selection Guide:
- Need a thinking partner? → chat (brainstorm ideas, get second opinions, validate approaches)
- Need deeper thinking? → thinkdeep (extends analysis, finds edge cases)
- Want to prevent "You're absolutely right!" responses? → challenge (challenges assumptions, encourages thoughtful re-evaluation)
- Need to break down complex projects? → planner (step-by-step planning, project structure, breaking down complex ideas)
- Need multiple perspectives? → consensus (get diverse expert opinions on proposals and decisions)
- Code needs review? → codereview (bugs, security, performance issues)
- Pre-commit validation? → precommit (validate git changes before committing)
- Something's broken? → debug (systematic investigation, step-by-step root cause analysis)
- Want to understand code? → analyze (architecture, patterns, dependencies)
- Code needs refactoring? → refactor (intelligent refactoring with decomposition focus)
- Need call-flow analysis? → tracer (generates prompts for execution tracing and dependency mapping)
- Need comprehensive tests? → testgen (generates test suites with edge cases)
- Security concerns? → secaudit (OWASP analysis, compliance evaluation, vulnerability assessment)
- Code needs documentation? → docgen (generates comprehensive documentation with complexity analysis)
- Which models are available? → listmodels (shows all configured providers and models)
- Server info? → version (version and configuration details)

Auto Mode: When DEFAULT_MODEL=auto, Claude automatically picks the best model for each task. You can override with: "Use flash for quick analysis" or "Use o3 to debug this".

Model Selection Examples:
- Complex architecture review → Claude picks Gemini Pro
- Quick formatting check → Claude picks Flash
- Logical debugging → Claude picks O3
- General explanations → Claude picks Flash for speed
- Local analysis → Claude picks your Ollama model

Pro Tip: Thinking modes (for Gemini models) control depth vs token cost. Use "minimal" or "low" for quick tasks, "high" or "max" for complex problems. Learn more

Tools Overview:
1. chat - Collaborative thinking and development conversations
2. thinkdeep - Extended reasoning and problem-solving
3. challenge - Critical challenge prompt, prevents You're absolutely right!
4. planner - Interactive sequential planning for complex projects
5. consensus - Multi-model consensus analysis with stance steering
6. codereview - Professional code review with severity levels
7. precommit - Validate git changes before committing
8. debug - Systematic investigation and debugging
9. analyze - General-purpose file and code analysis
10. refactor - Code refactoring with decomposition focus
11. tracer - Static code analysis prompt generator for call-flow mapping
12. testgen - Comprehensive test generation with edge case coverage
13. secaudit - Comprehensive security audit with OWASP Top 10 analysis
14. docgen - Comprehensive documentation generation with complexity analysis
15. listmodels - Display all available AI models organized by provider
16. version - Get server version and configuration

1. chat - General Development Chat & Collaborative Thinking

Your thinking partner for brainstorming, getting second opinions, and validating approaches. Perfect for technology comparisons, architecture discussions, and collaborative problem-solving.```
Chat with zen about the best approach for user authentication in my React app


**📖 Read More** - Detailed features, examples, and best practices

### 2. `thinkdeep` - Extended Reasoning Partner
Get a second opinion to augment Claude's own extended thinking. Uses specialized thinking models to challenge assumptions, identify edge cases, and provide alternative perspectives.```
The button won't animate when clicked, it seems something else is intercepting the clicks. Use thinkdeep with gemini pro after gathering related code and handing it the files
and find out what the root cause is

📖 Read More - Enhanced analysis capabilities and critical evaluation process

3. challenge - Critical Challenge Prompt

Encourages thoughtful reassessment of statements instead of automatic agreement, especially when you're wrong.
Wraps your input with instructions for critical thinking and honest analysis.```
challenge isn't adding this function to the base class a bad idea?


Normally, your favorite coding agent will enthusiastically reply with **“You’re absolutely right!”**—then proceed
to completely reverse the _correct_ strategy, without ever explaining why you're wrong.

  Example: Without vs With Zen

  **Without Zen:**

  **With Zen:**

**📖 Read More** - Challenge an approach or validate ideas with confidence

### 4. `planner` - Interactive Step-by-Step Planning
Break down complex projects or ideas into manageable, structured plans through step-by-step thinking.
Perfect for adding new features to an existing system, scaling up system design, migration strategies,
and architectural planning with branching and revision capabilities.

#### Pro Tip
Claude supports `sub-tasks` where it will spawn and run separate background tasks. You can ask Claude to
run Zen's planner with two separate ideas. Then when it's done, use Zen's `consensus` tool to pass the entire
plan and get expert perspective from two powerful AI models on which one to work on first! Like performing **AB** testing
in one-go without the wait!```
Create two separate sub-tasks: in one, using planner tool show me how to add natural language support
to my cooking app. In the other sub-task, use planner to plan how to add support for voice notes to my cooking app.
Once done, start a consensus by sharing both plans to o3 and flash to give me the final verdict. Which one do
I implement first?

📖 Read More - Step-by-step planning methodology and multi-session continuation

5. consensus - Multi-Model Perspective Gathering

Get diverse expert opinions from multiple AI models on technical proposals and decisions. Supports stance steering (for/against/neutral) and structured decision-making.```
Get a consensus with flash taking a supportive stance and gemini pro being critical to evaluate whether we should
migrate from REST to GraphQL for our API. I need a definitive answer.


**📖 Read More** - Multi-model orchestration and decision analysis

### 6. `codereview` - Professional Code Review
Comprehensive code analysis with prioritized feedback and severity levels. This workflow tool guides Claude through systematic investigation steps with forced pauses between each step to ensure thorough code examination, issue identification, and quality assessment before providing expert analysis.```
Perform a codereview with gemini pro especially the auth.py as I feel some of the code is bypassing security checks
and there may be more potential vulnerabilities. Find and share related code."

Tips:
* To avoid extra API costs, add do not use another model to run the entire codereview workflow locally.
* If you want to always consult an external model (recommended for code reviews) regardless of Claude’s confidence level in identifying issues, use must codereview using. This ensures the model is consulted even if Claude is certain about identified issues.

📖 Read More - Professional review workflow with step-by-step analysis

7. precommit - Pre-Commit Validation

Comprehensive review of staged/unstaged git changes across multiple repositories. This workflow tool guides Claude through systematic investigation of git changes, repository status, and file modifications across multiple steps before providing expert validation to ensure changes meet requirements and prevent regressions.```
Perform a thorough precommit with o3, we want to only highlight critical issues, no blockers, no regressions. I need
100% confidence this covers all the edge cases listed in FIX_ISSUE_ABC.md


#### Real Example:

* I asked claude (sonnet-4) to refactor some code to fix a known bug; gave it exact instructions
* The fix looked correct — and at a glance, so did the changes.
* More importantly, all existing unit tests passed (540+ unit tests) without a hitch. Zero regressions. Claude seemed confident and eager to commit. Everything looked perfect.

I then ran:```text
Run a precommit with o3 confirm our changes are sound and diffs are valid. Confirm this won't cause breakage or
regressions and codesmells are out

This passed the tracked and untracked changes / diffs and the related code to o3 for a thorough review via Zen's precommit tool.

Output:```text
⏺ Precommit Analysis Results - CRITICAL ISSUE FOUND

The O3 model has identified one HIGH-severity regression and several medium/low issues in our refactor. Here's the summary:

🚨 HIGH SEVERITY - MUST FIX BEFORE COMMIT

...


The reported issue was in fact a _very subtle bug_ that slipped through the quick glance — and a unit test for this exact case apparently
was missing (out of 540 existing tests!) - explains the zero reported regressions. The fix was ultimately simple, but the
fact Claude (and by extension, I) overlooked this, was a stark reminder: no number of eyeballs is ever enough. Fixed the
issue, ran `precommit` with o3 again and got:

 **RECOMMENDATION: PROCEED WITH COMMIT**

Nice! This is just one instance - take a look at another example here.

**Tips**:
* To avoid extra API costs, add `do not use another model` to run the entire precommit workflow locally.
* If you want to **always** consult an external model (recommended for pre-commit analysis) regardless of Claude’s confidence level in identifying issues, use `must precommit using `. This ensures the model is consulted even if Claude is _certain_ about identified issues.

**📖 Read More** - Multi-repository validation and change analysis

### 8. `debug` - Expert Debugging Assistant
Systematic investigation-guided debugging that walks Claude through step-by-step root cause analysis. This workflow
tool enforces a structured investigation process where Claude performs methodical code examination, evidence collection,
and hypothesis formation across multiple steps before receiving expert analysis from the selected AI model. When Claude's
confidence reaches **100% certainty** during the investigative workflow, expert analysis via another model is skipped to
save on tokens and cost, and Claude proceeds directly to fixing the issue.```
See logs under /Users/me/project/diagnostics.log and related code under the sync folder.
Logs show that sync works but sometimes it gets stuck and there are no errors displayed to
the user. Using zen's debug tool with gemini pro, find out why this is happening and what the root
cause is and its fix

Tips:
* To avoid extra API costs, add do not use another model to run the entire debugging workflow locally. This is recommended in most cases, as Claude typically identifies the root cause with high confidence by the end.
* If you want to always consult an external model regardless of Claude’s confidence level, use must debug using. This ensures the model is consulted even if Claude is certain about the issue.

When in doubt, you can always follow up with a new prompt and ask Claude to share its findings with another model:```text
Use continuation with thinkdeep, share details with o4-mini to find out what the best fix is for this


**📖 Read More** - Step-by-step investigation methodology with workflow enforcement

### 9. `analyze` - Smart File Analysis
General-purpose code understanding and exploration. This workflow tool guides Claude through systematic investigation of code structure, patterns, and architectural decisions across multiple steps, gathering comprehensive insights before providing expert analysis for architecture assessment, pattern detection, and strategic improvement recommendations.```
Use gemini to analyze main.py to understand how it works

📖 Read More - Comprehensive analysis workflow with step-by-step investigation

10. refactor - Intelligent Code Refactoring

Comprehensive refactoring analysis with top-down decomposition strategy. This workflow tool enforces systematic investigation of code smells, decomposition opportunities, and modernization possibilities across multiple steps, ensuring thorough analysis before providing expert refactoring recommendations with precise implementation guidance.```
Use gemini pro to decompose my_crazy_big_class.m into smaller extensions


**📖 Read More** - Workflow-driven refactoring with progressive analysis

### 11. `tracer` - Static Code Analysis Prompt Generator
Creates detailed analysis prompts for call-flow mapping and dependency tracing. Generates structured analysis requests for precision execution flow or dependency mapping.```
Use zen tracer to analyze how UserAuthManager.authenticate is used and why

📖 Read More - Prompt generation and analysis modes

12. testgen - Comprehensive Test Generation

Generates thorough test suites with edge case coverage based on existing code and test framework. This workflow tool guides Claude through systematic investigation of code functionality, critical paths, edge cases, and integration points across multiple steps before generating comprehensive tests with realistic failure mode analysis.Use zen to generate tests for User.login() method

…(正文过长,页面只展示前面部分,完整内容见仓库)

常见问题

6 问
答案由本站依据条目字段整理,不引用上游原文
Zen MCP Server E6de5294 怎么安装?

资料没有给出安装命令,请到官方仓库或文档按说明接入;这类条目多数是生态应用,或尚未发布可安装包。

Zen MCP Server E6de5294 怎么在 Claude Desktop、Claude 里配置?

它的传输方式未标注,在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,配置才会生效。

Zen MCP Server E6de5294 收费吗?

本站不代收费用,也没有标注具体价格。是否收费取决于服务本身和背后的接口额度,正式使用前按官方说明确认。

Zen MCP Server E6de5294 需要 API Key 吗?

资料标注需要API Key认证,使用前准备好对应凭据,并按最小权限配置。

Zen MCP Server E6de5294 支持哪些客户端?

资料标注支持 Claude Desktop、Claude;其他客户端只要能配置 MCP Server,也能用同样的命令或地址接入。

Zen MCP Server E6de5294 能提供哪些能力?

资料显示约 16 个工具;能力标签包括 AI协作、代码分析、Python、Claude。具体工具清单以官方文档为准。

热门专题

按标签挑服务器
全部专题 →

服务器信息以作者仓库与官方文档为准;分类、传输方式与鉴权方式的中文说明为中文整理,最近核验 2026-05-17。