Salesforce MCP Pro
A Model Context Protocol server for Salesforce development workflows, enabling org lookup, metadata deploy/retrieve, SOQL queries, Apex tests, and permission set assignment.
README
Salesforce MCP Pro
Salesforce MCP Pro is a high-performance Node 22 Model Context Protocol server for Salesforce development workflows, built as a faster and more controllable replacement for the standard Salesforce MCP server.
Why Salesforce MCP Pro
- Fast local startup and reliable operation on Windows and macOS without
npx @salesforce/mcp - More than 30 tools across DX operations, org discovery, metadata workflows, query/search, debugging, generation, and CRUD automation
- Hybrid architecture (Salesforce CLI + jsforce) for broader capability with predictable behavior
- Structured JSON responses designed for agent clients and repeatable team workflows
Requirements
- Node.js
>=22.19.0 - Salesforce CLI (
sf) installed and authenticated
Run
npm install
npm start
Cutover checklist
Use docs/cutover.md to migrate a project from @salesforce/mcp to Salesforce MCP Pro.
Cursor MCP config (portable Option A)
This option avoids machine-specific absolute paths and works across team members' computers when repos are cloned in a shared sibling layout.
Assumed layout:
.../<salesforce-project>.../SalesforceMCP
From <salesforce-project>/.cursor/mcp.json, use the relative path ../SalesforceMCP/src/index.js.
{
"mcpServers": {
"Salesforce MCP Pro": {
"command": "node",
"args": ["../SalesforceMCP/src/index.js"]
}
}
}
If you use Claude Desktop, use the same command and args under your mcpServers object.
Environment variables
SF_CLI_PATH(optional): absolute path tosfexecutable if not on PATH (normally not needed)ALLOWED_ORGS(optional): comma-separated org aliases/usernames allowedREAD_ONLY(optional):trueto block mutating toolsENFORCE_PROTECTED_ORG_CONFIRM(optional, defaulttrue): requireconfirm: truefor mutating tools when target org alias/username matches protected patternsPROTECTED_ORG_PATTERNS(optional): comma-separated values (default:prod,production,live)
Team prerequisites for portable config
- Node.js
>=22.19.0available on PATH (node -v) - Salesforce CLI available on PATH (
sf --version) - Salesforce CLI authenticated (
sf org list)
Tools
DX tools
Org and query tools
sf_get_usernamesf_list_all_orgssf_query_orgsf_query_org_paginatedsf_search_orgsf_execute_named_querysf_describe_object
Apex, LWC, and Flow generation tools
sf_execute_anonymous_apexsf_build_apex_cursor_querysf_execute_apex_cursor_querysf_generate_lwc_componentsf_generate_flow_definition
Debug and metadata tools
sf_list_debug_logssf_get_debug_logsf_enable_debug_logssf_disable_debug_logssf_deploy_metadatasf_deploy_metadata_validatesf_get_deploy_statussf_cancel_deploysf_list_metadatasf_retrieve_metadata
Testing and access tools
sf_run_apex_testsf_run_apex_test_suitesf_assign_permission_setsf_assign_permset_groupsf_remove_permission_setsf_remove_permset_groupsf_create_user
CRUD tools
sf_create_recordssf_update_recordssf_delete_recordssf_upsert_records
Local smoke test
- Start the server:
npm start - Verify it prints
Salesforce MCP Pro is running on stdioto stderr. - In Cursor, open MCP tools and confirm the
Salesforce MCP Proserver loads with tools.
Cutover
Use the cutover checklist in docs/cutover.md to migrate your project from @salesforce/mcp to Salesforce MCP Pro.
CRUD examples
See docs/crud-examples.md for ready-to-use payload examples for:
sf_create_recordssf_update_recordssf_delete_recordssf_upsert_records
Query and search examples
See docs/search-examples.md for ready-to-use payload examples for:
sf_query_orgsf_query_org_paginatedsf_search_orgsf_execute_named_querysf_describe_objectsf_build_apex_cursor_querysf_execute_apex_cursor_query
DX operations examples
See docs/dx-operations-examples.md for ready-to-use payload examples for:
sf_generate_lwc_componentsf_generate_flow_definitionsf_deploy_metadata_validatesf_get_deploy_statussf_cancel_deploysf_list_metadatasf_run_apex_test_suitesf_execute_anonymous_apexsf_remove_permission_setsf_remove_permset_groupsf_create_usersf_list_debug_logssf_get_debug_logsf_enable_debug_logssf_disable_debug_logs
Generation tools (sf_generate_lwc_component, sf_generate_flow_definition) also support optional deployment with deployAfterGenerate: true plus targetOrg, deployWait, and confirm (required for protected org aliases when safety guard is enabled).
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。