Sign in Book a demo
Back to Tools
Sales & CRM OAuth
Build Your Agent View Salesforce Docs
salesforce logo

Salesforce MCP Tools for AI Agents

Let an agent search Salesforce accounts, leads, contacts, and opportunities, create records, schedule tasks, and log calls.

Available Tools

17 tools available in this integration

Salesforce.ConvertLead

Convert a lead into a Contact (and optionally Account + Opportunity).

Salesforce.CreateContact

Create a contact in Salesforce.

Salesforce.CreateLead

Create a new lead in Salesforce.

Salesforce.CreateOpportunity

Create a new opportunity (deal) in Salesforce.

Salesforce.CreateTask

Create a task (follow-up, reminder, to-do) in Salesforce.

Salesforce.GetAccountDataById

Get the account with related info: contacts, leads, notes, calls, opportunities, tasks,

Salesforce.GetAccountDataByKeywords

Search for accounts in Salesforce and return them with related info: contacts, leads,

Salesforce.GetOpportunityById

Get a single opportunity with enriched related data: contact roles, line items,

Salesforce.ListMyTasks

List tasks assigned to you in Salesforce with optional filters.

Salesforce.LogACall

Log a completed phone call as an activity in Salesforce.

Salesforce.SearchContacts

Search for contacts in Salesforce with optional filters.

Salesforce.SearchLeads

Search for leads in Salesforce with optional filters.

Salesforce.SearchOpportunities

Search for opportunities (deals) in Salesforce with optional filters.

Salesforce.UpdateLead

Update fields on an existing lead.

Salesforce.UpdateOpportunity

Update fields on an existing opportunity.

Salesforce.UpdateTask

Update fields on an existing task.

Salesforce.WhoAmI

Get information about the currently authenticated Salesforce user.

Why agent-optimized tools matter

Built for intent, not API schemas

Ask the agent to "find the Acme opportunity, check its contacts and recent activity, and create the next follow-up task." Arcade helps it use the right Salesforce actions without guessing record IDs or API fields.

Up to 100x fewer tokens

Salesforce searches return focused records and related sales context instead of entire org exports. The agent can inspect the account, opportunity, contacts, and tasks it needs without loading a full CRM into the model.

Complete the workflow end to end

Find the account, inspect its opportunities, review related contacts and activity, create the follow-up task, and update the opportunity through one toolkit designed around the complete sales workflow.

Make actions predictable and controlled

Arcade tools return structured results and useful errors across users and runs. Teams can apply scope restrictions, redaction, and approval policies so agents follow the same controls every time.

Securing Salesforce

The more capabilities agents get, the more useful they are. However, more capabilities bring more risks of undesired effects such as data loss and exfiltration.

This is why Arcade provides you with contextual rules that dictate how and when a tool is allowed to run. Arcade’s contextual access lets you govern tool visibility and behavior at every stage of execution and control who can see a tool, what inputs are allowed, and what comes back.

Learn about Contextual Access Policies

For Salesforce, you can implement some of these policies:

Access when tools are listed for a user
  • Hide Salesforce.ConvertLead from agents unless the roles your IDP marks as sales managers are active, so an agent cannot turn an unqualified lead into a contact, account, or opportunity.
  • Restrict Salesforce.CreateOpportunity to agents operating under your pipeline-owner role, so unreviewed automation cannot create deals in your forecast.
  • Deny Salesforce.UpdateOpportunity to agents outside the roles your IDP marks as revenue administrators, so they cannot change deal stages, amounts, or close dates.
Pre-execution before each tool call
  • Require human approval before Salesforce.ConvertLead runs, so an agent cannot merge a lead into customer records without review.
  • Block Salesforce.CreateOpportunity when the proposed amount exceeds your approval threshold, so an agent cannot add a large unapproved deal to the pipeline.
  • Require human approval before Salesforce.LogACall runs when the activity includes a customer phone number or call notes, so private conversation details are not recorded without review.
Post-execution after the tool returns
  • Redact phone numbers, email addresses, and call notes from Salesforce.GetAccountDataById results before they reach the model, so account context does not expose direct contact details or private conversation content.
  • Strip line-item prices, discount details, and internal notes from Salesforce.GetOpportunityById results, so the model sees deal status without sensitive commercial terms.
  • Redact personal phone numbers and email addresses from Salesforce.SearchContacts results before the model sees them, so contact matching does not expose direct contact data.

How to use Salesforce with your agent

Pick your MCP client below and follow the setup walkthrough from authentication to your first Salesforce request. Each end user authorizes their own Salesforce account.

Your questions, answered

A Salesforce MCP server exposes Salesforce capabilities as structured tools that an AI agent can call through the Model Context Protocol. After a user authorizes Salesforce, the agent can search, read, create, update, and log permitted sales records and activities.

Arcade provides Salesforce through its own agent runtime rather than requiring you to build the OAuth and tool-calling layer yourself. It adds per-user authorization, agent-optimized actions, approval policies, redaction, and scope controls across supported clients.

Arcade agents can search leads, contacts, accounts, and opportunities; inspect related sales data; create and update leads, opportunities, contacts, and tasks; convert leads; log calls; and list the authenticated user's tasks.

Salesforce OAuth access must include the API access and refresh permissions needed for the workflow. Salesforce object- and field-level permissions still apply, so users need access to the records and actions the agent will use.

Yes. Each user authorizes their own Salesforce account through OAuth. Actions run under that user's authorization and Salesforce permissions, so the agent cannot access another user's account through their connection.

Yes. Contextual Access policies can require human approval before consequential Salesforce actions such as creating or updating sales records, converting a lead, logging a call, or changing a task. The call proceeds only after the policy's conditions are satisfied.

Arcade proxies Salesforce API calls and does not persist Salesforce content as part of normal tool execution. Credentials are brokered and stored encrypted at rest, and the model never receives them.

Any framework or client that can connect to Arcade's MCP gateway or tool APIs works, including LangChain, OpenAI Agents, CrewAI, Claude Desktop, Claude Code, Cursor, Visual Studio Code, Microsoft Copilot Studio, and GitHub Copilot.

Arcade refreshes Salesforce access while the user's authorization remains valid. If it is revoked, rotated, or can no longer be refreshed, calls stop until the user reconnects Salesforce.

Salesforce API limits, org quotas, and permissions still apply to calls made through Arcade. The available actions are individual record and activity operations rather than bulk data operations, so each search or write is subject to Salesforce's applicable limits.

Ready to actually ship your agent?

Connect your agent to Salesforce with managed OAuth, agent-optimized tools and approval controls for consequential actions.

Build Your Agent Read the Salesforce Quickstart