@restforge-dev/mcp-server
MCP server that exposes RESTForge capabilities to AI agents, enabling them to set up, configure, generate code, and manage RESTForge projects through natural language.
README
@restforge-dev/mcp-server
MCP (Model Context Protocol) server for the RESTForge framework. Exposes RESTForge capabilities to AI Agents (Claude Desktop, Cursor, Claude CLI, and other MCP clients) so agents can operate RESTForge through natural language without manually invoking CLI commands.
Requirements
- Node.js >= 18
- npm >= 9
- For full setup workflow: PostgreSQL / MySQL / Oracle / SQLite, RESTForge license key
Installation
npm install -g @restforge-dev/mcp-server
After installation, the restforge-mcp command is available in PATH.
Quick Start
1. Verify Install
echo '{"jsonrpc":"2.0","id":1,"method":"tools/list"}' | restforge-mcp
Output should list 29 tools across the health_*, setup_*, codegen_*, and runtime_* domains.
2. Register with MCP Client
Claude CLI (user scope, applies to all projects):
claude mcp add --transport stdio --scope user restforge -- restforge-mcp
Cursor (.cursor/mcp.json in project root):
{
"mcpServers": {
"restforge": {
"command": "restforge-mcp"
}
}
}
Claude Desktop (claude_desktop_config.json):
{
"mcpServers": {
"restforge": {
"command": "restforge-mcp"
}
}
}
3. Use via Natural Language
In your AI client chat, type prompts like:
Setup a new RESTForge project at
d:/projects/api-testwith PostgreSQL on localhost:5432, licenseXXXX-XXXX-XXXX-XXXX
Generate a CRUD endpoint for the
customertable
Run my RESTForge server (the agent generates a launcher script for the user to execute)
The agent orchestrates the appropriate tools to fulfill the request end-to-end.
Available Tools
29 tools organized by domain. AI agents call these via the MCP protocol; end users do not invoke them directly.
Health Domain (1 tool)
| Tool | Description |
|---|---|
health_ping |
Smoke test MCP transport. Returns pong + ISO timestamp + server version |
Setup Domain (9 tools)
| Tool | Description |
|---|---|
setup_create_folder |
Create a new project folder for RESTForge |
setup_install_package |
Install restforgejs into the project's node_modules via npm |
setup_init_config |
Generate skeleton config and sample payloads via restforge-cli init |
setup_write_env |
Write config/db-connection.env with license, server, and database settings |
setup_read_env |
Read current values from config/db-connection.env |
setup_update_env |
Update individual fields in config/db-connection.env |
setup_validate_config |
Validate license and connections to database, redis, and kafka |
setup_get_config_schema |
Get JSON schema of all 63 parameters available in db-connection.env |
setup_get_init_template |
Get raw db-connection.env template content |
Codegen Domain (13 tools)
| Tool | Description |
|---|---|
codegen_list_tables |
List all tables in the project's database (live introspection) |
codegen_describe_table |
Describe columns, primary key, and foreign keys of a specific table |
codegen_generate_payload |
Generate payload JSON from a database table |
codegen_validate_payload |
Validate payload JSON structure and constraints |
codegen_validate_dashboard_payload |
Validate dashboard payload structure |
codegen_diff_payload |
Diff payload JSON against the database schema |
codegen_sync_payload |
Sync payload JSON with the database schema |
codegen_create_endpoint |
Scaffold an endpoint module from a payload spec |
codegen_create_dashboard |
Scaffold a dashboard module from a payload spec |
codegen_validate_sql |
Validate a SELECT or WITH (CTE) SQL statement via EXPLAIN against the live database |
codegen_get_field_validation_catalog |
Get the field validation catalog (for grounding payload constraints) |
codegen_get_query_declarative_catalog |
Get the query declarative catalog (for grounding query JSON) |
codegen_get_dashboard_catalog |
Get the dashboard widget catalog (for grounding dashboard config) |
Runtime Domain (6 tools)
| Tool | Description |
|---|---|
runtime_detect_project |
Scan src/modules/*.js to list project names |
runtime_detect_config |
Scan config/*.env to list available config files |
runtime_validate_preflight |
Validate config + check PID file + check port availability before launch |
runtime_check_launcher_exists |
Check if launcher files (server-start.bat/.sh, ecosystem.config.js) exist in the project root |
runtime_generate_launcher |
Generate server-start.bat/.sh + server-stop.bat/.sh (and ecosystem.config.js for PM2 mode) |
runtime_check_status |
Detect if the server is running (host or PM2 mode) with optional HTTP health probe |
Runtime principle: AI agents never start, stop, or restart the server directly. The runtime tools only generate launcher scripts that the user executes themselves, so the running server lives independently of the AI session.
Compatibility
This MCP server works with any MCP client that supports the stdio transport, including but not limited to:
- Claude Desktop
- Claude CLI (Claude Code)
- Cursor
- Windsurf
- Cline (VS Code extension)
- Continue (VS Code/JetBrains extension)
- Zed
The model used (Claude, GPT, Gemini, etc.) depends on the client configuration. Tool selection accuracy is best with frontier models that have mature tool-calling support.
Repository
- Source: https://github.com/restforge/restforge-mcp
- Issues: https://github.com/restforge/restforge-mcp/issues
License
MIT — see LICENSE.md.
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。