Refactoring MCP
Exposes structured prompt generation tools for refactoring protocol and governance specifications via MCP, enabling teams to generate, validate, and retrieve standardized refactoring prompts through automation scripts and HTTP APIs.
README
Refactoring MCP — PRD
This project turns the Refactoring MCP prompt and governance specification templates into a dedicated MCP server so teams can host structured prompt generation via HTTP and automation scripts.
1. MCP Server Goal
- What is this project?
A dedicated MCP server (built on@modelcontextprotocol/sdk) that exposes MCP tools for generating protocol and governance prompt payloads following the Refactoring MCP standards. - What problem does it solve?
Documentation-only templates are hard to integrate; this MCP server exposes prompt builders via HTTP so downstream automation and review tooling can consume the specs. - Who is the primary user?
Platform engineers, MCP reviewers, or automation scripts needing consistent refactoring governance messaging via MCP workflows.
2. Success Definition
- How do we know this project is successful?
The npm package publishes, the MCP server starts, and prompts can be generated and stored via therefactoring_mcp_*tools. - What measurable outcome must be achieved?
npm testsucceeds,npm start(orrefactoring-mcp) brings the MCP service online, and at least one prompt record can be created through the MCP tools. - What does failure look like?
Prompt definitions remain stuck in Markdown without API access, or the automated validation/tests do not run.
3. Core MCP Server Action
- What is the single most important action a user performs?
Start the MCP server (vianpm startorrefactoring-mcp), host the API, and POST a protocol/governance prompt definition for downstream tooling. - What moment should make the user say “this works”?
A successfulPOST /api/promptsfollowed by aGET /api/prompts/:idthat returns the prompt text and metadata.
4. Non-Goals
- What will this project NOT attempt to do?
It does not persist prompts beyond in-memory, support authentication, or orchestrate remote agents. - What features are explicitly out of scope?
Any persistent database, full LLM orchestration, or advanced scheduling can be deferred to future MCPs.
5. Constraints
- Time constraints: Keep the MVP lean for npm publishing—no long-running refactors.
- Scope constraints: Single MCP server (Express + prompt builders) without sprawling CLI features.
- Resource constraints: Only trusted dependencies (Express, built-in Node modules) are permitted.
Validation Check
- Can the project goal be explained in one sentence? Yes — “Expose Refactoring MCP prompts via an Express API and CLI for automation-ready tooling.”
- Can success be evaluated without subjective judgment? Yes — endpoints respond successfully and tests pass.
6. MCP Server Usage
Installation
npm install -g refactoring-mcp(ornpm install refactoring-mcpfor local use).
Run (stdio)
refactoring-mcpornpm startboots the MCP server on port 3000 by default.- Use
--port <number>(or-p) or set thePORTenv var to change the listening port.
Example MCP Client Config
{
"command": "npx",
"args": ["-y", "refactoring-mcp"]
}
Tools
service_status→ return service metadata (name, version, uptime, timestamp).refactoring_mcp_list_documents→ list the guidance documents that describe this MCP.refactoring_mcp_get_document→ fetch a specific guidance document by key.refactoring_mcp_get_bundle→ download all guidance documents in one bundle.refactoring_mcp_generate_prompt→ build a protocol or governance prompt, validate the schema, and store the payload.refactoring_mcp_list_prompts→ inspect saved prompt metadata and timestamps.refactoring_mcp_get_prompt→ retrieve a stored prompt byid.
Documents
spec→docs/spec-prompts.md(protocol & governance spec templates).trd→TRD.md(technical boundaries for the MCP server).mcp_server→docs/mcp-server.md(server responsibilities and customization steps).agents→examples/AGENTS.md(agent instructions tied to this MCP).trd_questions→prompts/trd.questions.md(prompt template for drafting the TRD).
Tool Overview
service_statuskeeps MCP agents aware of the running version and uptime of the server.refactoring_mcp_*tools encapsulate prompt generation, doc discovery, and persisted prompt storage so other agents can consume the MCP output.
MCP Server Runtime
refactoring-mcp(installed globally) ornpm startstarts the MCP server so agents can access prompt specs.- Override the port with
--port,-p, or thePORTenvironment variable.
Testing & Validation
npm testrunstsc(build) and thennode --test ./dist/test/*.test.js, covering the prompt builders, store, and validators.
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。