Salesforce CRM

Enable AI agents to query and modify Salesforce data using SOQL, manage records, and access object metadata

The Salesforce MCP server enables AI agents to interact with Salesforce CRM data, including querying records with SOQL, creating and updating objects, and accessing metadata.

Overview

This connector provides tools for:

• SOQL Queries - Query any Salesforce object with full SOQL support
• Record Management - Create, update, and delete records
• Object Metadata - Discover available objects and their fields
• Bulk Operations - Handle multiple records efficiently

Integration Options

Option A: Salesforce Hosted MCP (Coming Soon)

Salesforce is developing an official hosted MCP server, currently in beta with general availability expected in February 2026.

Sign up for the beta at the Salesforce Developer Blog.

Option B: Deploy to Heroku

Deploy a community MCP server to Heroku:

After deployment, configure your Salesforce credentials:

heroku config:set SF_INSTANCE_URL=https://your-instance.salesforce.com -a your-app-name
heroku config:set SF_CLIENT_ID=your-connected-app-client-id -a your-app-name
heroku config:set SF_CLIENT_SECRET=your-connected-app-client-secret -a your-app-name
heroku config:set SF_USERNAME=your-salesforce-username -a your-app-name
heroku config:set SF_PASSWORD=your-salesforce-password -a your-app-name
heroku config:set SF_SECURITY_TOKEN=your-security-token -a your-app-name

Register with Heroku AI

heroku ai:mcp:servers:add salesforce-mcp \
  --app your-inference-app \
  --server-app your-salesforce-mcp-app

Available Tools

Tool	Description
query	Execute SOQL queries against Salesforce
describe_object	Get metadata about a Salesforce object (fields, types, relationships)
create	Create a new record in any Salesforce object
update	Update an existing record by ID
delete	Delete a record by ID

Tool Details

query

Execute SOQL (Salesforce Object Query Language) queries to retrieve data.

Parameters:

• query (string, required) - The SOQL query to execute

Example queries:

-- Get recent opportunities
SELECT Id, Name, Amount, StageName, CloseDate
FROM Opportunity
WHERE CloseDate = THIS_QUARTER
ORDER BY Amount DESC

-- Find contacts at a company
SELECT Id, FirstName, LastName, Email, Phone
FROM Contact
WHERE Account.Name = 'Acme Corp'

-- Get open cases
SELECT Id, CaseNumber, Subject, Status, Priority
FROM Case
WHERE IsClosed = false
ORDER BY CreatedDate DESC

describe_object

Get metadata about a Salesforce object including fields, data types, and relationships.

Parameters:

• object_name (string, required) - API name of the Salesforce object (e.g., Account, Contact, Opportunity)

create

Create a new record in Salesforce.

Parameters:

• object_name (string, required) - API name of the object
• fields (object, required) - Field values for the new record

Example:

{
  "object_name": "Contact",
  "fields": {
    "FirstName": "Jane",
    "LastName": "Doe",
    "Email": "[e-mail]",
    "AccountId": "001xx000003DGbYAAW"
  }
}

update

Update an existing record.

Parameters:

• object_name (string, required) - API name of the object
• record_id (string, required) - Salesforce record ID
• fields (object, required) - Fields to update

delete

Delete a record from Salesforce.

Parameters:

• object_name (string, required) - API name of the object
• record_id (string, required) - Salesforce record ID to delete

Using with Heroku AI Agents

Python

import os
from openai import OpenAI

client = OpenAI(
    base_url=os.getenv("INFERENCE_URL") + "/v1",
    api_key=os.getenv("INFERENCE_KEY")
)

response = client.chat.completions.create(
    model=os.getenv("INFERENCE_MODEL_ID"),
    messages=[
        {"role": "user", "content": "Find all opportunities closing this month over $50,000"}
    ],
    extra_body={
        "heroku": {
            "mcp_servers": ["salesforce-mcp"]
        }
    }
)

print(response.choices[0].message.content)

cURL

curl $INFERENCE_URL/v1/agents/heroku \
  -H "Authorization: Bearer $INFERENCE_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "'$INFERENCE_MODEL_ID'",
    "messages": [
      {"role": "user", "content": "Find all opportunities closing this month over $50,000"}
    ],
    "mcp_servers": ["salesforce-mcp"]
  }'

Configuration

Variable	Description	Required
SF_INSTANCE_URL	Your Salesforce instance URL	Yes
SF_CLIENT_ID	Connected App consumer key	Yes
SF_CLIENT_SECRET	Connected App consumer secret	Yes
SF_USERNAME	Salesforce username	Yes
SF_PASSWORD	Salesforce password	Yes
SF_SECURITY_TOKEN	Security token (append to password if required)	Depends

Setting Up a Connected App

1. Go to Setup → App Manager → New Connected App
2. Enable OAuth Settings
3. Add OAuth scopes: api, refresh_token, offline_access
4. Save and note the Consumer Key and Consumer Secret
5. Configure your Heroku app with the credentials

See Salesforce Connected Apps Guide for detailed instructions.

Example Use Cases

Sales Pipeline

"Show me all opportunities in the Negotiation stage with close dates in the next 30 days"

Account Research

"Get the contact information for all decision makers at Acme Corp"

Data Entry

"Create a new lead for John Smith at TechCorp, email [e-mail], phone [phone]"

Reporting

"Summarize our win rate by industry for Q4"

Case Management

"Find all high-priority cases opened this week that haven't been assigned"

Security Considerations

• Principle of Least Privilege - Use a Salesforce user with minimal required permissions
• Field-Level Security - The MCP server respects Salesforce field-level security
• API Limits - Monitor your Salesforce API usage limits
• Audit Logging - All operations are logged in Salesforce’s audit trail

Additional Resources

• Salesforce REST API Documentation
• SOQL Reference
• Heroku MCP Integration Guide
• Heroku Agents API