Open-source B2B and marketplace ecommerce; new stores include an MCP server for AI agents.
Sfcc Dev MCP
概览
作者提供的说明(英文原文)
MCP server for Salesforce B2C Commerce Cloud development assistance
PopNx MCP 的三大支柱
这三项永远不会合并成一个分数:意见、测量和人气是不同的东西。
- 用户评价i暂无评价评价功能将在后续阶段与账户功能一起开放
- 我们的基础检查i1 项中通过 1 项依据公开信息检查;尚未扫描代码
- 使用情况i尚未统计★ –
它能帮你做什么
我们还没有为它写通俗的介绍。下面是我们从注册中心读取到的事实。
- 连接方式: 在你的电脑上运行
- 传输方式: stdio
- 分类: 网店与支付
安装
选择你使用的 AI 应用,然后复制设置。
安装之前
- 基础检查:1 项中通过 1 项
- 在你的电脑上运行,拥有你的用户账户的权限(文件、网络、应用)。
- 注册中心没有列出需要填写的密钥。
你需要
- npm已安装 Node.js(用于 npx 命令)
你用的是哪个应用?
在上方选择你的应用。我们会记住它,下次按钮会显示“安装到 …”。
打开 Claude Desktop → Settings → Developer → Edit Config,把下面的内容粘贴到 claude_desktop_config.json
{
"mcpServers": {
"sfcc-dev": {
"command": "npx",
"args": [
"-y",
"sfcc-dev-mcp"
]
}
}
}在终端中运行这条命令
claude mcp add sfcc-dev -- npx -y sfcc-dev-mcp或者自己复制设置
粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"sfcc-dev": {
"command": "npx",
"args": [
"-y",
"sfcc-dev-mcp"
]
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
或者自己复制设置
粘贴到项目中的 .vscode/mcp.json
{
"servers": {
"sfcc-dev": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"sfcc-dev-mcp"
]
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
这个服务器在你的电脑上运行,而 ChatGPT 只能连接互联网上的服务器。请选择其他应用。
你的选择通过 Cookie 保存在这台设备上,无需账户。
这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同,请以 README 为准。
它能做什么
它在你的电脑上运行;安装后,工具列表会显示在你的 AI 应用中。
基础检查i
我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月7日
- ✓ 通过
作者与仓库一致
注册名称已通过 GitHub 账户 taurgis 验证,该账户拥有此仓库
- … 尚未检查
仍在维护
我们还没有从 GitHub 读取这个仓库
- … 尚未检查
有许可证
我们还没有从 GitHub 读取这个仓库
- – 不适用
端点有响应
它在你的电脑上运行,没有互联网端点
- – 不适用
使用 OAuth 登录
在你的电脑上运行,改用本地设置
- … 尚未检查
工具描述中没有隐藏指令
它在你的电脑上运行;我们只能读取互联网端点的工具列表
与“网店与支付”中的其他服务器比较i
- 星标数多于此分类中其他 1754 个服务器的 0%
- 发布到注册中心的时间比此分类中其他 1754 个服务器的 5% 更新
- 基础检查结果等于或优于此分类中其他 1754 个服务器的 100%
与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。
常见问题
Sfcc Dev MCP 安全吗?
它在 1 项适用的基础检查中通过了 1 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。
它需要哪些密钥或代码?
注册中心没有列出必需的密钥,不过有些服务器会在首次使用时要求你登录。
它在我的电脑上运行还是在云端?
它在你的电脑上运行。你的 AI 应用会在需要时启动它,数据不会经过作者的服务器(除非它自己调用在线服务)。
它还在维护吗?
GitHub 仓库最后更新于–,最新版本(1.1.0)于7个月前进入注册中心。
哪些应用可以使用它?
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。你的电脑上还需要有 Node.js。
有什么可以替代它?
同一分类中有:Spree Commerce: open-source B2B and marketplace ecommerce, Stripe MCP, MCP Cn Commerce
替代选择
同一分类中其他作者的服务器,按星标数排序。
如何看卡片:★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i
Stripe MCP
MCP server integrating with Stripe - tools for customers, products, payments, and more.
MCP Cn Commerce
8 read-only MCP platform entries; support is operation-specific; merchant live acceptance pending
作者的 README
打开 README
按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。
SFCC Development MCP Server
An AI-powered Model Context Protocol (MCP) server that provides comprehensive access to Salesforce B2C Commerce Cloud development tools, documentation, and runtime diagnostics.
✨ Key Features
- 🔍 Complete SFCC Documentation Access - Search and explore all SFCC API classes and methods
- 🏗️ SFRA Documentation - Enhanced access to Storefront Reference Architecture documentation
- 🧱 ISML Template Reference - Complete ISML element documentation with examples and usage guidance
- 📊 Log Analysis Tools - Real-time error monitoring, debugging, and job log analysis for SFCC instances
- ⚙️ System Object Definitions - Explore custom attributes and site preferences
- 🧪 Script Debugger - Execute and inspect script-debugger endpoints in credentialed mode, including custom trigger URLs/paths for non-default storefront routes
- 🚀 Cartridge Generation - Automated cartridge structure creation with workspace-bound path safety (writes stay inside workspace roots, or current working directory fallback when roots are unavailable; home-directory fallback is blocked)
- 🧩 Agent Skill Bootstrap - Install or merge AGENTS.md and bundled skills into the current project or a temp directory for AI assistants
- ✅ Tool Argument Validation - Runtime schema validation enforces required fields, type checks, enum constraints, integer/numeric bounds, and strict unknown-key checks for object schemas (top-level and nested) before handler execution
- ⏱️ MCP Progress + Cancellation - Tool calls honor request cancellation signals and emit out-of-band
notifications/progressupdates when clients provide aprogressToken
🚀 Quick Start
Option 1: Documentation-Only Mode (No SFCC credentials needed)
{
"mcpServers": { "sfcc-dev": { "command": "npx", "args": ["sfcc-dev-mcp"] } } }
Option 2: Full Mode (With SFCC credentials for log and job analysis)
{
"mcpServers": { "sfcc-dev": { "command": "npx", "args": ["sfcc-dev-mcp", "--dw-json", "/path/to/your/dw.json"] } } }
Create a dw.json file with your SFCC credentials. You can use either auth mode (or both):
- Basic auth:
username+password - OAuth:
client-id+client-secret - Optional storefront auth (for script debugger trigger on Basic-Auth storefronts):
storefrontUsername+storefrontPassword
{
"hostname": "your-instance.sandbox.us01.dx.commercecloud.salesforce.com", "username": "your-username", "password": "your-password", "storefrontUsername": "your-storefront-basic-user", "storefrontPassword": "your-storefront-basic-password", "client-id": "your-client-id", "client-secret": "your-client-secret" }
At least one complete credential pair is required when hostname is set. If credentials are provided, hostname is also required.
Option 3: Auto-Discovery (Recommended for VS Code users)
Simply open a VS Code workspace that contains a dw.json file - the server will automatically discover and use it:
{
"mcpServers": { "sfcc-dev": { "command": "npx", "args": ["sfcc-dev-mcp"] } } }
🔧 Configuration Discovery Priority
The server discovers SFCC credentials in this order (highest priority first):
| Priority | Source | Description |
|---|---|---|
| 1 | --dw-json CLI parameter | Explicit path to dw.json file |
| 2 | Environment variables | SFCC_HOSTNAME, SFCC_USERNAME, SFCC_PASSWORD, SFCC_CLIENT_ID, SFCC_CLIENT_SECRET |
| 3 | MCP workspace roots | Automatically discovers dw.json in your VS Code workspace folder(s), and refreshes when the client sends notifications/roots/list_changed |
Note: The server no longer searches the current working directory by default, as MCP servers often start with
cwdset to the user's home directory. The MCP workspace roots mechanism provides reliable project context.
🎯 Operating Modes
| Mode | Tools Available | SFCC Credentials Required |
|---|---|---|
| Documentation-Only | 18 tools | ❌ No |
| Full Mode | 40 tools | ✅ Yes |
Documentation-Only Mode
Perfect for learning and development, no SFCC instance required:
- Complete SFCC API documentation (5 tools)
- SFRA documentation (5 tools)
- ISML template documentation (5 tools)
- Cartridge generation (1 tool, writes constrained to workspace roots/cwd)
- Agent instruction bootstrap (2 tools) to copy/merge AGENTS.md and skills, or disable future prompts
Full Mode
Complete development experience with live SFCC instance access:
- All documentation-only features (18 tools)
- Real-time log analysis and job logs (13 tools)
- System object definitions (6 tools)
- Code version management (2 tools)
- Script debugger operations (1 tool)
🏗️ Architecture Overview
This server is built around a capability-gated, modular handler architecture that cleanly separates tool routing from domain logic:
Core Layers
- Tool Schemas (
src/core/tool-schemas/): Modular, category-based tool definitions (documentation, SFRA, ISML, logs, job logs, system objects, cartridge, code versions, agent instructions, script debugger). Re-exported viatool-definitions.ts. - Server Orchestration Modules (
src/core/server-tool-catalog.ts,src/core/server-tool-call-lifecycle.ts,src/core/server-workspace-discovery.ts): Keepsserver.tsfocused by extracting capability-aware tool catalog logic,tools/calllifecycle (progress/cancellation/preflight), and workspace roots reconfiguration flow. - Tool Argument Validator (
src/core/tool-argument-validator.ts): Enforces runtime argument shape at the MCP boundary for all tools (required fields, primitive/object/array types, enum checks, integer/numeric ranges, string patterns/length, and strict unknown-key checks for object schemas at top-level and nested levels) before tool dispatch. - OCAPI Query Coverage (
src/core/tool-schemas/shared-schemas.ts): Shared search schemas includetext_query,term_query,bool_query,filtered_query, andmatch_all_queryso MCP boundary validation aligns with supported OCAPI query patterns. - Handlers (
src/core/handlers/): Each category has a handler extending a common base for timing, structured logging, and error normalization, with config-driven wiring viaConfiguredClientHandlerto reduce repetitive boilerplate (e.g.log-handler,docs-handler,isml-handler,system-object-handler). - Clients (
src/clients/): Encapsulate domain operations (OCAPI, SFRA docs, ISML docs, modular log analysis, script debugger, cartridge generation, agent-instruction sync). Handlers delegate to these so orchestration and computation remain separate. - Services (
src/services/): Dependency-injected abstractions for filesystem and path operations — improves testability and isolates side effects. - Modular Log System (
src/clients/logs/): Reader (range/tail optimization), discovery, processor (line → structured entry), analyzer (patterns & health), formatter (human output) for maintainable evolution. - Configuration Factory (
src/config/configuration-factory.ts): Determines capabilities (canAccessLogs,canAccessOCAPI) based on provided credentials and filters exposed tools accordingly (principle of least privilege). - Shared Credential Validation (
src/config/credential-validation.ts): Centralizes auth-pair completeness and hostname-format validation for bothdw.jsonloading and runtime configuration creation. - Call-time Capability Guarding (
src/core/server.ts): Rejects execution of tools that are unavailable in the current mode, so hidden tools are not callable via directtools/callrequests. - Call Lifecycle Signals (
src/core/server.ts):tools/callhandling supports cancellation via request abort signals and emits best-effort progress notifications when the caller provides_meta.progressToken. - Tool Error Sanitization (
src/core/tool-error-response.ts): Sanitizes upstream execution errors before returning MCP tool responses, reducing accidental leakage of backend payload details. - Runtime WebDAV Verification (
src/core/server.ts): For OAuth-only configurations (client-id/client-secretwithoutusername/password), log/job-log/script-debugger tool exposure is gated by a one-time WebDAV capability probe to avoid false-positive tool availability. - CLI Option Helpers (
src/config/cli-options.ts): Centralizes command-line parsing and environment credential detection for predictable startup behavior. - Shared Path Security Policy (
src/config/path-security-policy.ts): Reuses allow/block path rules across workspace-root discovery and securedw.jsonloading. - Shared Abort Utility (
src/utils/abort-utils.ts): Centralized timeout and abort-signal composition used by HTTP and debugger clients for consistent cancellation semantics and timer cleanup.
Why This Matters
- Extensibility: Adding a new tool usually means adding a schema + minimal handler logic (or a new handler if a new domain).
- Security: Tools that require credentials are never exposed when capability flags are false.
- Testability: Unit tests target clients & modules; integration/MCP tests validate handler routing and response structure.
- Performance: Tail log reads + lightweight caching (
cache.ts,log-cache.ts) reduce unnecessary I/O.
Adding a New Tool (High-Level)
- Add schema to the appropriate file in
src/core/tool-schemas/(or create new file for new category). - Export new schema from
src/core/tool-schemas/index.tsif adding a new file. - Implement domain logic in a client/service (avoid bloating handlers).
- Extend an existing handler or create a new one if it's a new category.
- (Only for a new category) register the new handler inside
registerHandlers()inserver.ts. - Discover actual response shape with
npx aegis querybefore writing tests. - Add Jest unit tests + YAML MCP tests (docs vs full mode if credentials required).
- Update documentation (AGENTS.md + README counts if changed).
For a deeper internal view, see the Development Guide in the docs site.
🤖 AI Interface Setup
Choose your preferred AI assistant:
| Interface | Best For | Setup Guide |
|---|---|---|
| Claude Desktop | Multi-turn conversations, debugging | 📖 Setup Guide |
| GitHub Copilot | VS Code integration, inline suggestions | 📖 Setup Guide |
| Cursor | Modern AI-powered editor | 📖 Setup Guide |
📦 Installation
Using npx (Recommended)
Tip: Add
-y(or--yes) to suppress the interactive prompt npx shows before downloading a package. This prevents AI clients (Claude Desktop, Copilot, Cursor) from hanging waiting for confirmation.
# Test the server
npx -y sfcc-dev-mcp
Use with your configuration
npx -y sfcc-dev-mcp --dw-json /path/to/your/dw.json
Global Installation
npm install -g sfcc-dev-mcp
sfcc-dev-mcp --dw-json /path/to/your/dw.json
🐛 Debug Mode & Logging
Enable Debug Logging
# Enable debug mode for detailed logging
npx -y sfcc-dev-mcp --debug
Or with configuration file
npx -y sfcc-dev-mcp --dw-json /path/to/your/dw.json --debug
--debug accepts true/false, 1/0, or yes/no. Invalid values fail fast with a clear error message.
Log File Locations
The server writes logs to your system's temporary directory:
- macOS: `/var/folders/{user-id}/T/sfcc-mcp-
只显示了开头部分;其余内容请在仓库中阅读