Scaffold

MCP开发工具

提供方 Beer-Bears

开发工具 stdio · 本地进程 会话登录 社区 / 第三方 CursorCodexZed
查看接入方式

Scaffold is a specialized RAG system that transforms entire codebase into a living knowledge graph, capturing the deep structural relationships between files, classes, and functions. This provides AI agents with the crucial context to accurately understand, maintain, and reason a

一句话选型结论

按资料字段整理

Scaffold 属于开发工具类 MCP 服务器,stdio · 本地进程,会话登录,可直接接入 Cursor、Codex、Zed,有可直接执行的安装命令。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stdio · 本地进程 鉴权会话登录 部署方式本地部署
GitHub 星标 未标注 仓库热度,无公开仓库时显示未标注
工具数 未标注 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
工具数量未标注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

graph rag rag llm knowledge graph code analysis generative ai

支持客户端

适用场景

graph ragragllm

安装与配置

stdio · 本地进程 会话登录
客户端在本机拉起子进程,数据不出本机
docker run -d \\

这条命令只是入口,具体启动参数请按官方文档补齐,不要直接照搬到别的服务器。

  1. 1
    准备运行环境

    stdio 服务在本地起进程,需要先装好 Cross-platform,再按安装命令拉取依赖。

  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

    连接成功后随便调一个只读工具,确认能返回结果,再接入正式流程。

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

  • 有现成的安装命令,不用自己找入口,照着跑就能起服务。
  • 已适配 Cursor、Codex、Zed,换客户端不用重新研究配置。

先想清楚的地方

  • stdio 方式要求本地装好运行时(如 Node.js),换机器要重新装一遍。
  • 使用前要准备好会话登录对应的凭据,并按最小权限配置。
  • 资料没标注工具数量,实际能力要装好后确认。
  • 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 Cursor、Codex 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

  • 完全不想在本地装环境的人。
  • 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
Scaffold当前 未标注 stdio · 本地进程 本地部署 有 未标注
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Beer-Bears上游标注的维护方 Beer-Bears
分类开发工具本站按上游分类展示
连接方式stdio · 本地进程客户端在本机拉起子进程,数据不出本机
鉴权方式会话登录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 · Cross-platform本地安装时用到的包标识与运行时
部署类型本地部署上游给出的部署形态
最近核验未标注上游最后核验时间

详细介绍

7.6 KB
来自项目仓库与上游目录的原文,未做删改

Scaffold is a specialized RAG (Retrieval-Augmented Generation) system designed to revolutionize how development teams interact with large codebases. Born from real-world frustrations with traditional documentation and AI-assisted development, Scaffold provides the structural foundation AI agents need to effectively construct, maintain, and repair complex software projects.

The Challenge

Modern development teams face three critical problems:

  1. Documentation Decay: Maintaining accurate and up-to-date technical documentation requires unsustainable manual effort.
  2. AI Context Blindness: LLMs lack awareness of project-specific architecture and business logic, requiring inefficient manual context provisioning.
  3. Knowledge Fragmentation: Critical system understanding exists only in tribal knowledge that's lost when team members leave.

Our Solution

Scaffold transforms your source code into a living knowledge graph stored in a graph database. This creates an intelligent context layer that:

  • Captures structural relationships between code entities.
  • Maintains both vector and graph representations of your codebase.
  • Enables precise context injection for LLMs and AI agents.
  • Supports construction, maintenance, and refactoring workflows.

> Like its physical namesake, Scaffold provides the temporary support structure needed to build something great - then disappears when the work is done.


Getting Started

There are two primary ways to run Scaffold. Choose the one that best fits your needs.

Option 1: Run with Pre-built Docker Image (Recommended for Users)

> This is the fastest method to get Scaffold running. It uses the official pre-built image from the GitHub Container Registry and does not require you to clone the source code repository. You only need to create two configuration files.

1. Prepare Your Project Directory

Create a new folder for your project setup.```bash
mkdir my-scaffold-server
cd my-scaffold-server


#### 2. Create Configuration Files
In the `my-scaffold-server` directory, create the following two files.

**`docker-compose.yaml`:**```yaml
services:
  scaffold-mcp:
    image: ghcr.io/beer-bears/scaffold:latest
    container_name: scaffold-mcp-prod
    env_file:
      - .env
    tty: true
    ports:
      - "8000:8080"
    depends_on:
      - neo4j
    volumes:
      - ./codebase:/app/codebase

  chromadb:
    image: chromadb/chroma:1.0.13
    container_name: scaffold-chromadb
    restart: unless-stopped
    volumes:
      - chroma_data:/data

  neo4j:
    image: neo4j:5
    container_name: scaffold-neo4j
    restart: unless-stopped
    environment:
      NEO4J_AUTH: "${NEO4J_USER:-neo4j}/${NEO4J_PASSWORD:-password}"
    volumes:
      - neo4j_data:/data
    ports:
      - "7474:7474"
      - "7687:7687"

volumes:
  chroma_data:
  neo4j_data:

.env:```dotenv

ChromaDB Settings

CHROMA_SERVER_HOST=chromadb
CHROMA_SERVER_PORT=8000
CHROMA_COLLECTION_NAME=scaffold_data

Neo4j Credentials

NEO4J_USER=neo4j
NEO4J_PASSWORD=password
NEO4J_URI=bolt://neo4j:password@neo4j:7687

Absolute path to your codebase

PROJECT_PATH=<ABSOLUTE_PATH_TO_YOUR_CODEBASE>


#### 3. Run the Application
Start all services using Docker Compose.```bash
docker-compose up -d

Scaffold will now start and begin analyzing the code of your codebase.

Option 2: Build from Source (For Developers)

> This method is for developers who have cloned the repository and want to build the Docker image locally. This is ideal for contributing to Scaffold or making custom modifications, or just run all containers at once in self-hosting mode.

1. Set Up The Project

First, clone the repository and navigate into the project directory.```bash
git clone https://github.com/Beer-Bears/scaffold.git
cd scaffold

Next, create your environment file from the example provided.```bash
cp .env.example .env

2. Add Your Codebase

Place the Python project you want to analyze into the codebase directory.```bash

Create the directory if it doesn't exist

mkdir -p codebase

Copy your project files into it

cp -r /path/to/your/python/project/* ./codebase/

Alternatively, you can edit the `.env` file and set the `PROJECT_PATH` variable to the absolute path of your project on your host machine.

#### 3. Run the Application
Start the entire application stack using Docker Compose. The `--build` flag will compile your local source code into a new Docker image.```bash
docker-compose up --build -d

Interact with Scaffold

Once the containers are running (using either method), you can interact with the system.

A. Configure your MCP Client (e.g., Cursor)

Add the Scaffold server to your client's mcp.json file.```json
{
"mcpServers": {
"scaffold-mcp": {
"url": "http://localhost:8000/mcp"
}
}
}


#### B. Explore the Knowledge Graph
Access the Neo4j web UI to visually explore the graph of your codebase. Use the credentials from your `.env` file (default: `neo4j` / `password`).
**URL:** http://localhost:7474/

#### C. Send a Direct API Request
You can also test the MCP endpoint directly using `curl`.```bash
curl -N -X POST http://localhost:8000/mcp/ \
  -H &quot;Content-Type: application/json&quot; \
  -H &quot;Accept: application/json, text/event-stream&quot; \
  -d &#x27;{
    &quot;jsonrpc&quot;: &quot;2.0&quot;,
    &quot;id&quot;: &quot;1&quot;,
    &quot;method&quot;: &quot;tools/call&quot;,
    &quot;params&quot;: {
      &quot;name&quot;: &quot;get_code_entity_information&quot;,
      &quot;arguments&quot;: {
        &quot;entity_name&quot;: &quot;MyClassName&quot;
      }
    }
  }&#x27;

> You also can check out mcp description and get docker cmd configuration


How It Works

High-Level Architecture

View Architecture Schema

Usecase & Interface Diagrams

View More Diagrams

Usecase Schema

Interfaces Schema

Project Structure```

.
├── docs
│ ├── img # Static Images
│ └── research # Research reports
└── src
├── core # RAG Context Fetching Algorithms
├── database # Graph/Vector Database Logic
├── generator # Abstract Tree Generator
├── mcp # MCP Interface
└── parsers # AST Parcers
```

FAQ

What is RAG (Retrieval-Augmented Generation)?

RAG (Retrieval-Augmented Generation) is a technique that enhances large language models (LLMs) by:

  1. Retrieving relevant information from external knowledge sources
  2. Augmenting the LLM's context with this retrieved information
  3. Generating more accurate, context-aware responses

Unlike traditional LLMs that rely solely on their training data, RAG systems access up-to-date project-specific information and reduce hallucinations by grounding responses in actual codebase context.

How does Graph RAG work?

Graph RAG extends traditional RAG by representing knowledge as interconnected entities in a graph database. This allows the system to understand and retrieve not just chunks of text, but also the structural relationships between them (e.g., this function calls another function, this class inherits from another class). This structural context is invaluable for complex software engineering tasks.

Resources

  • What is Retrieval-Augmented Generation (RAG)?
  • GraphRAG vs. Traditional RAG: Higher Accuracy & Insight with LLM
  • What is MCP? Integrate AI Agents with Databases & APIs

Contributing

Scaffold is an open-source project and we welcome contributions from the community! Whether it's reporting a bug, discussing features, or submitting code, your help is valued.

To get started, please read our Contributing Guidelines for details on how to set up your development environment, run tests, and submit a pull request.

License

This project is licensed under the MIT License. See the LICENSE file for details.

常见问题

8 问
答案由本站依据条目字段整理,不引用上游原文
Scaffold 怎么安装?

在客户端按下面的安装命令接入:docker run -d \\。装完在客户端配置里引用同一个包或端点。

Scaffold 怎么在 Cursor、Codex、Zed 里配置?

它的传输方式是stdio · 本地进程,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。

Scaffold 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。

Scaffold 需要 API Key 吗?

资料标注需要会话登录认证,使用前准备好对应凭据,并按最小权限配置。

Scaffold 支持哪些客户端?

资料标注支持 Cursor、Codex、Zed;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。

Scaffold 能提供哪些能力?

资料显示能力标签包括 graph rag、rag、llm、knowledge graph、code analysis、generative ai。具体工具清单以官方文档为准。

Scaffold 需要什么运行环境?

运行时为 Cross-platform;属于本地进程,需要在本机装好运行环境。

Scaffold 是本地还是远程?

它是本地进程(stdio),由客户端在本机拉起。传输方式为stdio · 本地进程。

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 未标注。