Dev Guru
An AI-powered code consultation server that routes programming queries to specific AI models based on requested expertise levels. It enables users to receive structured feedback on debugging, architectural decisions, and code reviews from Gemini, Claude, or GPT.
README
<div align="center">
<img src="guru.png" alt="Dev Guru" width="300"/>
🧘 Dev Guru
Your AI-powered code consultation MCP server.
When you're stuck, afraid, or just lazy to ask for help — Dev Guru is here.
</div>
💡 What is Dev Guru?
Dev Guru is a specialized MCP (Model Context Protocol) server that acts as an on-demand senior code consultant for AI agents. It routes coding problems to the most suitable AI model based on the requested expertise level, providing structured, actionable feedback.
Think of it as a second brain for your AI agent — a guru it can consult when facing tough coding decisions.
🎯 Use Cases
| Scenario | How Dev Guru Helps |
|---|---|
| 🐛 Debugging Complex Issues | Your agent is stuck on a tricky bug. It calls Dev Guru with the context and gets expert-level reasoning and suggestions. |
| 🏗️ Architecture Decisions | Unsure about a design pattern? Dev Guru analyzes your code structure and recommends the best approach. |
| 🔄 Code Review on Demand | Submit code for review and get structured feedback with a thinking process and concrete suggestions. |
| 🤔 Validating Reasoning | Your agent has an idea but isn't confident. Dev Guru validates the reasoning and either confirms or corrects the approach. |
| ⚡ Multi-Model Leverage | Automatically routes to Gemini, Claude, or GPT based on the complexity level — getting the right model for the right job. |
✨ Features
- 🧠 Expert-based Routing — Automatically selects the best AI model for the task:
novice→ Gemini (fast, efficient)medium→ Claude (balanced, analytical)expert→ OpenAI GPT (deep reasoning)
- 🔀 OpenRouter Fallback — If a primary API key is missing, seamlessly falls back to OpenRouter
- 🎛️ Configurable Models — Choose exactly which model to use per level via environment variables
- ⚡ FastMCP Core — High-performance MCP server implementation
- 📦 Skill Management API — Dynamic skill installation and management via REST
- 🐳 Docker Ready — Multi-stage build with
uvfor efficient containerized deployments - 🧩 Agno Framework — Leverages Agno for agent orchestration and structured outputs
🚀 Quick Start
Prerequisites
- Python 3.12+
uv(recommended)- At least one API key: Gemini, Anthropic, OpenAI, or OpenRouter
Installation
# Clone the repository
git clone https://github.com/your-user/dev-guru.git
cd dev-guru
# Create your environment file
cp .env.example .env
# Edit .env with your API keys
# Install dependencies
uv sync
Running
# Start the full API + MCP server
uv run python main.py
Docker
docker compose up --build
⚙️ Configuration
Environment Variables
| Variable | Description | Default |
|---|---|---|
GEMINI_API_KEY |
Google Gemini API key | — |
ANTHROPIC_API_KEY |
Anthropic Claude API key | — |
OPENAI_API_KEY |
OpenAI API key | — |
OPENROUTER_API_KEY |
OpenRouter API key (universal fallback) | — |
API_KEY |
Optional API key to protect REST and MCP endpoints | — |
NOVICE_MODEL |
Model ID for novice level | gemini-3.1-pro-preview |
MEDIUM_MODEL |
Model ID for medium level | claude-opus-4.6 |
ADVANCED_MODEL |
Model ID for expert level | gpt-5.3-codex |
PORT |
Server port | 8000 |
DEBUG |
Debug mode | true |
Tip: You only need an
OPENROUTER_API_KEYto use all three levels — it acts as a universal fallback for any missing provider key.
🔌 MCP Configuration
Add Dev Guru to your MCP client (Claude Desktop, Cursor, etc.):
{
"mcpServers": {
"dev-guru": {
"command": "uv",
"args": [
"--directory",
"/path/to/dev-guru",
"run",
"python",
"src/server.py"
]
}
}
}
📡 API Endpoints
Skill Management
| Method | Endpoint | Description |
|---|---|---|
GET |
/skills |
List all loaded skills |
GET |
/skills/{name} |
Get details of a specific skill |
POST |
/skills |
Install a skill (URL or base64 zip) |
POST |
/skills/upload |
Install a skill via file upload |
DELETE |
/skills/{name} |
Delete a skill |
MCP Tool
| Tool | Parameters | Description |
|---|---|---|
call_guru |
level, technologies, context, thinking |
Consult the guru about a coding problem |
🧪 Testing
PYTHONPATH=. uv run pytest
🏗️ Architecture
graph LR
A[AI Agent] -->|MCP Protocol| B[Dev Guru Server]
B -->|novice| C[Gemini]
B -->|medium| D[Claude]
B -->|expert| E[GPT-5.3-codex]
B -.->|fallback| F[OpenRouter]
F --> C
F --> D
F --> E
<div align="center">
Built with 🧘 by devs, for devs.
</div>
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。