Salesforce AI Agent MCP Server
Enables Claude to interact with Jira, Salesforce, and n8n through MCP tools, allowing automated reading of Jira stories, deploying Salesforce metadata, and triggering n8n workflows.
README
Salesforce AI Agent MCP Server
A production-grade Model Context Protocol (MCP) server that gives Claude direct, structured access to Jira Cloud, Salesforce (Tooling + Metadata APIs), and n8n workflows.
Built for the Salesforce Developer AI Agent project — an automation pipeline that reads Jira stories, interprets Salesforce configuration requirements, and deploys metadata (custom fields, validation rules, etc.) directly into Salesforce.
Architecture
Claude Desktop / Claude Code / VS Code
│ (stdio or SSE)
▼
salesforce-ai-agent-mcp
┌───────────────────────────┐
│ Tools (12 total) │
│ ├── Jira (4 tools) │──► Jira Cloud REST API v3
│ ├── Salesforce (6 tools) │──► Salesforce Tooling + Metadata APIs
│ └── n8n (2 tools) │──► n8n Webhook + REST API
└───────────────────────────┘
Prerequisites
- Node.js ≥ 18
- npm ≥ 9
- A Jira Cloud account with API token
- A Salesforce org with a Connected App (OAuth 2.0)
- An n8n instance (optional, for workflow orchestration)
Installation
cd salesforce-ai-agent-mcp
npm install
npm run build
Configuration
Copy .env.example to .env and fill in your credentials:
cp .env.example .env
Jira
| Variable | Description |
|---|---|
JIRA_BASE_URL |
e.g. https://your-org.atlassian.net |
JIRA_EMAIL |
Your Atlassian account email |
JIRA_API_TOKEN |
Generate at Atlassian API Tokens |
Salesforce
You need a Connected App with the Username-Password OAuth flow enabled.
- In Salesforce Setup → App Manager → New Connected App
- Enable OAuth settings
- Add scope:
api,refresh_token - Copy Consumer Key →
SF_CLIENT_ID - Copy Consumer Secret →
SF_CLIENT_SECRET
| Variable | Description |
|---|---|
SF_LOGIN_URL |
https://login.salesforce.com (prod) or https://test.salesforce.com (sandbox) |
SF_CLIENT_ID |
Connected App Consumer Key |
SF_CLIENT_SECRET |
Connected App Consumer Secret |
SF_USERNAME |
Your Salesforce username |
SF_PASSWORD |
Your Salesforce password |
SF_SECURITY_TOKEN |
Reset at Setup → Personal Information → Reset Security Token |
SF_API_VERSION |
API version, e.g. v61.0 (default) |
n8n
| Variable | Description |
|---|---|
N8N_BASE_URL |
Your n8n instance URL, e.g. https://your-n8n.com |
N8N_API_KEY |
Generate at n8n Settings → API → Create API Key |
Usage
stdio mode (Claude Desktop / Claude Code / VS Code)
npm start
# or for development:
npm run dev
SSE mode (for remote/n8n integration)
npm run start:sse
# or for development:
npm run dev:sse
Endpoints available in SSE mode:
GET http://localhost:3000/sse— clients connect herePOST http://localhost:3000/messages?sessionId=<id>— clients send messages hereGET http://localhost:3000/health— liveness probe
Override the port:
MCP_PORT=8080 npm run start:sse
Connecting to Claude Desktop
Add the following to your claude_desktop_config.json:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"salesforce-ai-agent": {
"command": "node",
"args": ["/absolute/path/to/salesforce-ai-agent-mcp/dist/index.js"],
"env": {
"JIRA_BASE_URL": "https://your-org.atlassian.net",
"JIRA_EMAIL": "you@company.com",
"JIRA_API_TOKEN": "your_token",
"SF_LOGIN_URL": "https://login.salesforce.com",
"SF_CLIENT_ID": "your_client_id",
"SF_CLIENT_SECRET": "your_client_secret",
"SF_USERNAME": "you@yourorg.com",
"SF_PASSWORD": "yourpassword",
"SF_SECURITY_TOKEN": "yourSecurityToken",
"N8N_BASE_URL": "https://your-n8n.com",
"N8N_API_KEY": "your_n8n_api_key"
}
}
}
}
Connecting to Claude Code (VS Code Extension)
Add to your VS Code settings.json (or via the MCP extension UI):
{
"mcp.servers": {
"salesforce-ai-agent": {
"command": "node",
"args": ["${workspaceFolder}/salesforce-ai-agent-mcp/dist/index.js"],
"env": {
"JIRA_BASE_URL": "https://your-org.atlassian.net",
"JIRA_EMAIL": "you@company.com",
"JIRA_API_TOKEN": "your_token",
"SF_LOGIN_URL": "https://login.salesforce.com",
"SF_CLIENT_ID": "your_client_id",
"SF_CLIENT_SECRET": "your_client_secret",
"SF_USERNAME": "you@yourorg.com",
"SF_PASSWORD": "yourpassword",
"SF_SECURITY_TOKEN": "yourSecurityToken",
"N8N_BASE_URL": "https://your-n8n.com",
"N8N_API_KEY": "your_n8n_api_key"
}
}
}
}
Alternatively, the mcp.json file at the root of this project is compatible with VS Code's MCP extension and will be auto-detected if placed in your workspace root.
Available Tools (12 total)
Jira Tools
jira_get_story
Fetch a Jira story by issue key. Returns summary, description, acceptance criteria, status, labels, and all custom fields.
issueKey: "SFDC-123"
jira_update_story_status
Transition a Jira story to a new workflow status.
issueKey: "SFDC-123"
targetStatus: "In Progress" # must match an available transition
jira_post_comment
Post a comment to a Jira story (used by the AI agent to report results or ask for clarification).
issueKey: "SFDC-123"
body: "Deployed Customer_Tier__c field to Account. Deployment ID: 0Af..."
jira_search_stories
JQL-based story search.
jql: "label = \"sf-config\" AND status = \"To Do\" AND project = SFDC"
maxResults: 25
Salesforce Tools
salesforce_get_object_fields
Describe all fields on a Salesforce object.
objectName: "Account"
salesforce_create_custom_field
Create a custom field via the Tooling API.
objectName: "Account"
fieldLabel: "Customer Tier"
fieldApiName: "Customer_Tier" # __c added automatically
fieldType: "Picklist"
picklistValues: ["Platinum", "Gold", "Silver", "Bronze"]
description: "Tier classification for account segmentation"
Lookup field example:
objectName: "Case"
fieldLabel: "Related Contract"
fieldApiName: "Related_Contract"
fieldType: "Lookup"
referenceTo: "Contract"
salesforce_create_validation_rule
Create a validation rule via the Tooling API.
objectName: "Account"
ruleName: "Require_Phone_For_Hot_Leads"
errorConditionFormula: "AND(Rating = \"Hot\", ISBLANK(Phone))"
errorMessage: "Phone number is required for Hot-rated accounts"
errorDisplayField: "Phone"
active: true
salesforce_deploy_metadata
Trigger a metadata deployment from a base64-encoded ZIP.
zipFile: "<base64-encoded-zip>"
checkOnly: false
testLevel: "RunLocalTests"
rollbackOnError: true
salesforce_get_deployment_status
Poll a deployment's status.
deploymentId: "0AfXXXXXXXXXXXXX"
salesforce_query
Execute SOQL for validation and verification.
soql: "SELECT Id, Name, Customer_Tier__c FROM Account WHERE Rating = 'Hot' LIMIT 10"
n8n Tools
n8n_trigger_workflow
Trigger an n8n workflow via webhook URL.
webhookUrl: "https://your-n8n.com/webhook/abc123"
payload: {
"issueKey": "SFDC-123",
"environment": "sandbox",
"triggeredBy": "claude"
}
n8n_get_execution_status
Check the status of an n8n execution.
executionId: "12345"
Example AI Agent Workflow
You: "Process Jira story SFDC-456 and deploy the Salesforce configuration"
Claude:
1. jira_get_story(issueKey: "SFDC-456")
→ reads requirements: "Add Customer_Tier picklist to Account"
2. salesforce_get_object_fields(objectName: "Account")
→ confirms Customer_Tier__c doesn't exist yet
3. salesforce_create_custom_field(
objectName: "Account",
fieldLabel: "Customer Tier",
fieldApiName: "Customer_Tier",
fieldType: "Picklist",
picklistValues: ["Platinum", "Gold", "Silver"]
)
→ creates the field
4. salesforce_query(soql: "SELECT QualifiedApiName FROM FieldDefinition WHERE EntityDefinition.QualifiedApiName = 'Account' AND QualifiedApiName = 'Customer_Tier__c'")
→ verifies the field was created
5. jira_post_comment(
issueKey: "SFDC-456",
body: "✅ Customer_Tier__c picklist field created on Account object with values: Platinum, Gold, Silver."
)
6. jira_update_story_status(issueKey: "SFDC-456", targetStatus: "Done")
Logging
All logs are written as structured JSON to stderr so they never interfere with the MCP stdio transport.
Control the log level via environment variable:
LOG_LEVEL=debug npm start # debug | info | warn | error
Development
# Type-check only
npm run typecheck
# Build
npm run build
# Dev with hot reload (stdio)
npm run dev
# Dev with hot reload (SSE)
npm run dev:sse
Project Structure
salesforce-ai-agent-mcp/
├── src/
│ ├── index.ts # Entry point — stdio & SSE transport setup
│ ├── server.ts # McpServer creation & tool registration
│ ├── tools/
│ │ ├── jira.ts # Jira tool definitions (4 tools)
│ │ ├── salesforce.ts # Salesforce tool definitions (6 tools)
│ │ └── n8n.ts # n8n tool definitions (2 tools)
│ ├── clients/
│ │ ├── jiraClient.ts # Jira REST API v3 client
│ │ ├── salesforceClient.ts # Salesforce Tooling/Metadata API + OAuth
│ │ ├── n8nClient.ts # n8n webhook + REST API client
│ │ └── logger.ts # Structured JSON logger (stderr)
│ └── types/
│ └── index.ts # Shared TypeScript types
├── .env.example # All required environment variables
├── mcp.json # VS Code MCP extension manifest
├── package.json
├── tsconfig.json
└── README.md
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.