Salesforce MCP Server
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Salesforce MCP Server List all opportunities with amount > 10000"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Salesforce MCP Server
Salesforce Mcp Server
A highly customizable Model Context Protocol (MCP) server for integrating Salesforce APIs with GenAI applications.
Features
Comprehensive Salesforce API Coverage:
SOQL queries with automatic pagination support
SOSL (Salesforce Object Search Language) searches
Record CRUD operations (Create, Read, Update, Delete)
Metadata API access for object descriptions
Bulk API operations for large data volumes
Apex code execution capabilities
Reports API with filtering support
Organization limits and API usage monitoring
Flexible Authentication:
OAuth 2.0 Web Server Flow
OAuth 2.0 JWT Bearer Flow
Username-Password Flow
Connected App support
Enterprise-Ready:
Multi-org support
Rate limiting and retry logic
Comprehensive error handling
Audit logging
Field-level security respect
Related MCP server: MCP Salesforce Connector
Installation
pip install salesforce-mcp-serverOr install from source:
git clone https://github.com/asklokesh/salesforce-mcp-server.git
cd salesforce-mcp-server
pip install -e .Configuration
Create a .env file or set environment variables:
# Salesforce Credentials
SALESFORCE_USERNAME=your_username@company.com
SALESFORCE_PASSWORD=your_password
SALESFORCE_SECURITY_TOKEN=your_security_token
SALESFORCE_DOMAIN=login # or test, or your custom domain
# OR use OAuth
SALESFORCE_CLIENT_ID=your_connected_app_client_id
SALESFORCE_CLIENT_SECRET=your_connected_app_client_secret
SALESFORCE_REDIRECT_URI=http://localhost:8080/callback
# Optional Settings
SALESFORCE_API_VERSION=59.0
SALESFORCE_SANDBOX=false
SALESFORCE_MAX_RETRIES=3
SALESFORCE_TIMEOUT=30Requirements
Python 3.10 or newer
MCP SDK 1.27+
Quick Start
Run the server
The server is built on the MCP Python SDK's FastMCP API and speaks the stdio transport. Install the package, configure credentials via environment variables (see Configuration), and launch with:
salesforce-mcp
# or equivalently
python -m salesforce_mcp.serverProgrammatic usage
from salesforce_mcp import mcp
# `mcp` is a configured `FastMCP` instance with every Salesforce tool
# already registered. Run it over stdio:
if __name__ == "__main__":
mcp.run(transport="stdio")Claude Desktop Configuration
Add to your Claude Desktop config:
{
"mcpServers": {
"salesforce": {
"command": "salesforce-mcp",
"env": {
"SALESFORCE_USERNAME": "your_username@company.com",
"SALESFORCE_PASSWORD": "your_password",
"SALESFORCE_SECURITY_TOKEN": "your_security_token"
}
}
}
}Available Tools
1. Query Records
Execute SOQL queries to retrieve data:
{
"tool": "salesforce_query",
"arguments": {
"query": "SELECT Id, Name, Email FROM Contact WHERE LastModifiedDate = TODAY",
"include_deleted": false
}
}2. Get Record
Retrieve a specific record by ID:
{
"tool": "salesforce_get_record",
"arguments": {
"object_type": "Account",
"record_id": "001XX000003DHPh",
"fields": ["Name", "Industry", "AnnualRevenue"]
}
}3. Create Record
Create new records:
{
"tool": "salesforce_create_record",
"arguments": {
"object_type": "Contact",
"data": {
"FirstName": "John",
"LastName": "Doe",
"Email": "john.doe@example.com",
"AccountId": "001XX000003DHPh"
}
}
}4. Update Record
Update existing records:
{
"tool": "salesforce_update_record",
"arguments": {
"object_type": "Contact",
"record_id": "003XX000004TMM2",
"data": {
"Title": "Senior Developer",
"Department": "Engineering"
}
}
}5. Delete Record
Delete records:
{
"tool": "salesforce_delete_record",
"arguments": {
"object_type": "Contact",
"record_id": "003XX000004TMM2"
}
}6. Describe Object
Get metadata about Salesforce objects:
{
"tool": "salesforce_describe_object",
"arguments": {
"object_type": "Account"
}
}7. Bulk Operations
Handle large data volumes:
{
"tool": "salesforce_bulk_create",
"arguments": {
"object_type": "Contact",
"records": [
{"FirstName": "Jane", "LastName": "Smith", "Email": "jane@example.com"},
{"FirstName": "Bob", "LastName": "Johnson", "Email": "bob@example.com"}
],
"batch_size": 200
}
}8. Execute Apex
Run Apex code:
{
"tool": "salesforce_execute_apex",
"arguments": {
"apex_body": "System.debug('Hello from Apex!');"
}
}9. Search Records (SOSL)
Search across multiple objects:
{
"tool": "salesforce_search",
"arguments": {
"search_query": "FIND {John} IN NAME FIELDS RETURNING Contact(Id, Name, Email), Account(Id, Name)"
}
}10. Query Pagination
Handle large query results:
{
"tool": "salesforce_query_more",
"arguments": {
"next_records_url": "/services/data/v59.0/query/01gxx000000002ABC-2000"
}
}11. Organization Limits
Get API limits and usage:
{
"tool": "salesforce_limits",
"arguments": {}
}12. Run Reports
Execute Salesforce reports:
{
"tool": "salesforce_run_report",
"arguments": {
"report_id": "00O1i000004Gzr0EAC",
"filters": {
"reportMetadata": {
"reportFilters": [
{
"column": "CREATED_DATE",
"operator": "greaterThan",
"value": "2023-01-01"
}
]
}
}
}
}Advanced Configuration
Multi-Org Support
Additional orgs are configured by setting prefixed environment variables.
For example, to expose a sandbox org alongside the default:
SALESFORCE_USERNAME=prod@company.com
SALESFORCE_PASSWORD=prod_password
SALESFORCE_SECURITY_TOKEN=prod_token
SALESFORCE_SANDBOX_USERNAME=sandbox@company.com.sandbox
SALESFORCE_SANDBOX_PASSWORD=sandbox_password
SALESFORCE_SANDBOX_SECURITY_TOKEN=sandbox_token
SALESFORCE_SANDBOX_DOMAIN=testEvery tool accepts an optional org argument; omitting it targets the
default org defined by SALESFORCE_DEFAULT_ORG (defaulting to "default").
Authentication classes
The salesforce_mcp.auth module exposes UsernamePasswordAuth, OAuth2Auth,
and JWTAuth for use when embedding the underlying SalesforceClient in
your own code:
from salesforce_mcp import JWTAuth, SalesforceClient
auth = JWTAuth(
client_id="your_client_id",
username="your_username",
private_key_file="path/to/private_key.pem",
sandbox=False,
)
client = SalesforceClient(auth=auth)Rate Limiting
Rate limiting is enabled by default and tuned via environment variables:
SALESFORCE_RATE_LIMIT_ENABLED=true
SALESFORCE_RATE_LIMIT_REQUESTS_PER_SECOND=10
SALESFORCE_RATE_LIMIT_BURST_SIZE=20Integration Examples
See the examples/ directory for complete integration examples:
basic_usage.py- Simple queries and CRUD operationsbulk_operations.py- Handling large data volumesgenai_integration.py- Integration with GenAI APIsmulti_org.py- Managing multiple Salesforce orgsoauth_flow.py- OAuth authentication setup
Error Handling
The server provides detailed error information:
try:
result = server.execute_tool("salesforce_query", {
"query": "SELECT InvalidField FROM Account"
})
except SalesforceError as e:
print(f"Salesforce error: {e.error_code} - {e.message}")
print(f"Fields available: {e.available_fields}")Security Best Practices
Never commit credentials - Use environment variables or secure vaults
Use OAuth when possible - More secure than username/password
Implement field-level security - Respect Salesforce permissions
Enable audit logging - Track all API operations
Use IP restrictions - Limit access to known IP ranges
Contributing
Contributions are welcome! Please read our contributing guidelines and submit pull requests.
License
MIT License - see LICENSE file for details
Available Tools
13 toolssalesforce_bulk_createB
Insert multiple records using the Salesforce Bulk API 2.0.
Inputs: records (required), object_type (required), org, batch_size. Returns: object
salesforce_create_recordC
Create a new Salesforce record of the given object type.
Inputs: data (required), object_type (required), org. Returns: object
salesforce_delete_recordC
Delete a Salesforce record by ID.
Inputs: record_id (required), object_type (required), org. Returns: object
salesforce_describe_objectC
Return field/metadata information for a Salesforce object.
Inputs: object_type (required), org. Returns: object
salesforce_execute_apexC
Execute anonymous Apex code.
Inputs: apex_body (required), org. Returns: object
salesforce_get_recordB
Retrieve a Salesforce record by ID, optionally limiting returned fields.
Inputs: record_id (required), object_type (required), org, fields. Returns: object
salesforce_limitsC
Return the Salesforce organization's API limits and usage.
Inputs: org. Returns: object
salesforce_list_objectsB
List all Salesforce objects visible to the authenticated user.
Inputs: org. Returns: object
salesforce_queryA
Execute a SOQL query. Set include_deleted=True to search IsDeleted records.
Inputs: query (required), org, include_deleted. Returns: object
salesforce_query_moreA
Retrieve the next page of a paginated SOQL query.
Inputs: next_records_url (required), org. Returns: object
salesforce_run_reportC
Run a Salesforce analytics report.
Inputs: report_id (required), org, filters. Returns: object
salesforce_searchC
Execute a SOSL (Salesforce Object Search Language) search.
Inputs: search_query (required), org. Returns: object
salesforce_update_recordB
Update fields on an existing Salesforce record.
Inputs: data (required), record_id (required), object_type (required), org. Returns: object
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 13 tools
Each tool targets a distinct operation: SOQL query, SOSL search, record CRUD by ID, bulk create, pagination, describe, list objects, run report, execute Apex, and limits. The boundaries are clear; even query and query_more are distinguished by pagination.
The naming pattern is largely consistent with salesforce_ prefix and verb_noun structure (e.g., create_record, update_record). However, 'salesforce_limits' is a noun-only exception, and 'salesforce_query_more' uses an adverb qualifier, which slightly deviates from the strict verb_noun pattern.
Thirteen tools is well-scoped for a Salesforce MCP server, covering a broad range of operations without bloat. The count sits comfortably in the ideal 3-15 range and each tool serves a distinct purpose.
The tool surface covers core CRUD, querying, searching, pagination, object metadata, bulk insert, report execution, anonymous Apex, and API limits. Missing are bulk update/delete and upsert operations, but the coverage is solid for most typical Salesforce automation scenarios.
Maintenance
Related MCP Connectors
MCP server unifying ERPs, CRMs, APIs and knowledge base for Claude, ChatGPT and Gemini.
MCP server connecting AI agents to 100+ apps (Gmail, Slack, Notion, GitHub) via one-click OAuth.
MCP server that lets AI assistants use all OneSchema features exposed via the public API.
Hosted MCP server for AI-driven data ops. Create apps, manage schemas, and CRUD structured data.
Related MCP Servers
- AlicenseAqualityAmaintenanceAn MCP server implementation that integrates Claude with Salesforce, enabling natural language interactions with Salesforce data and metadata for querying, modifying, and managing objects and records.152,670 npm164MIT
- AlicenseBqualityCmaintenanceA Model Context Protocol server that enables LLMs to interact with Salesforce data through SOQL queries, SOSL searches, and various API operations including record management.14306 PyPI179MIT
- AlicenseAqualityDmaintenanceAn MCP server implementation that integrates Claude with Salesforce, enabling natural language interactions with Salesforce data and metadata for querying, modifying, and managing objects and records.755 npm15MIT
- AlicenseAqualityDmaintenanceAn MCP server implementation that integrates Claude with Salesforce, enabling natural language interactions with Salesforce data and metadata.855 npmMIT