Deslopify
A universal MCP server that acts as a code quality gate for AI assistants, providing pre-generation guidance, post-generation review, and root cause analysis to improve code quality.
README
Deslopify
Stop AI slop at the source.
A universal MCP server that acts as a code quality gate for AI assistants. Deslopify gives AI coding tools pre-generation guidance, post-generation review, and root cause analysis — so the code you get is idiomatic, generic, tested, and actually fixes the right problem.
🌐 deslopify.co.uk · 📦 npm · 🐙 GitHub
The Problem
AI assistants generate code that technically works but is often:
- Hyper-specific — hardcoded to one type or use case instead of generic
- Symptom-patching — fixes the reported bug but ignores the root cause
- Non-idiomatic — uses patterns from the wrong language
- Untested — no error handling, no edge cases, no tests
- Hallucinated — references APIs that don't exist in the target version
There's no feedback loop telling the AI "that's slop, do better." Deslopify is that feedback loop.
How It Works
Deslopify is an MCP server. Any MCP-compatible AI client connects to it and gets access to tools, prompts, and resources that enforce code quality.
┌──────────────────────────────┐
│ AI Client │
│ (Claude, Cursor, Windsurf) │
└──────────┬───────────────────┘
│ MCP
┌──────────▼───────────────────┐
│ Deslopify MCP Server │
│ │
│ ┌─────────┐ ┌────────────┐ │
│ │ Tools │ │ Prompts │ │
│ └─────────┘ └────────────┘ │
│ ┌─────────┐ ┌────────────┐ │
│ │ Rules │ │ Resources │ │
│ └─────────┘ └────────────┘ │
└──────────────────────────────┘
Quick Start
npx deslopify
Or install globally:
npm install -g deslopify
Claude Desktop (SSE)
Add to your claude_desktop_config.json:
{
"mcpServers": {
"deslopify": {
"url": "http://optiplex:7381/sse"
}
}
}
Claude Code / Codex CLI (Streamable HTTP)
claude mcp add deslopify --url http://optiplex:7381/mcp
Remote (once DNS propagates)
claude mcp add deslopify --url https://deslopify.co.uk/mcp
Tools
| Tool | When to Call | What It Does |
|---|---|---|
get_guidance |
Before writing code | Returns language idioms, pitfalls, and architecture notes |
review_code |
After writing code | Scores code quality, lists issues with severity and fix suggestions |
validate_approach |
Before fixing a bug | Detects symptom-only fixes and traces root causes |
suggest_tests |
After writing code | Generates test cases with framework-specific skeletons |
check_dependencies |
When adding packages | Flags abandoned, overkill, or deprecated dependencies |
check_api_exists |
When using unfamiliar APIs | Verifies APIs exist in the target language version |
Supported Languages
- C++ (C++17/20/23) — RAII, smart pointers, const correctness, ranges, move semantics
- C# (.NET 6/7/8+) — using/IDisposable, async/await, nullable refs, LINQ, DI, records
- Python (3.10+) — context managers, type hints, dataclasses, pathlib, generators
- Java (17/21+) — try-with-resources, records, Optional, Stream API, switch expressions
- TypeScript (5.x+) — strict mode, no any, discriminated unions, proper async, zod validation
Prompts
Deslopify includes 4 prompt templates accessible via the MCP prompt menu:
- Deslopify: Before Writing Code — pre-generation guidance
- Deslopify: After Writing Code — post-generation review loop
- Deslopify: Before Fixing a Bug — root cause validation
- Deslopify: Full Quality Workflow — complete guidance → review → test pipeline
Configuration
| Environment Variable | Default | Description |
|---|---|---|
DESLOPIFY_STRICTNESS |
medium |
low / medium / high |
DESLOPIFY_TRANSPORT |
stdio |
stdio / sse |
DESLOPIFY_PORT |
3000 |
Port for SSE transport |
ANTHROPIC_API_KEY |
— | Required for medium/high strictness AI layer |
Docker
docker run -d --name deslopify \
-e DESLOPIFY_TRANSPORT=sse \
-e DESLOPIFY_PORT=3000 \
-p 3000:3000 \
ghcr.io/veroxsity/deslopify:latest
Contributing
Rules are the heart of Deslopify. To add a new rule:
- Fork the repo
- Add your rule to the appropriate JSON file in
rules/ - Include: id, severity, description, detection, fix, bad example, good example
- Open a PR
See DESIGN.md for the full architecture and rule schema.
License
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。