Skip to main content
Glama
Salesforce MCP Pro

Salesforce MCP Pro

by adamcmyhre

Salesforce MCP Pro

Salesforce MCP Pro is a high-performance Node 22 MCP server for Salesforce development workflows. It is designed as a faster, more controllable replacement for the standard Salesforce MCP server, with broad DX and data coverage.

Why Salesforce MCP Pro

  • Fast startup and reliable local execution on Windows/macOS

  • More than 75 Salesforce-focused tools across DX, data, and UI automation domains

  • Hybrid backend (Salesforce CLI + jsforce) for broad coverage and practical performance

  • Built-in operational safety controls for production-like orgs

  • Structured stderr observability with per-tool request IDs, timing, org alias, and error classification

Related MCP server: Salesforce MCP Server

Capability Overview

Salesforce MCP Pro currently provides tool categories for:

  • Org discovery and environment resolution

  • Agentforce lifecycle and programmatic preview tooling

  • Tooling API discovery, metadata inspection, and dependency impact analysis

  • Metadata API deploy/retrieve lifecycle operations

  • 2GP packaging lifecycle operations

  • SOQL, SOSL, Apex Cursors, Named Query API, DML operations

  • Apex testing, anonymous execution and async job operations (schedule/batch)

  • Chatter collaboration tooling for record feeds and group updates

  • Nebula Logger wrappers for namespace-aware log and tag queries

  • LWC and Flow generation

  • Multi-Framework support: GraphQL queries and UIBundle scaffold/list/build/deploy workflows

  • Create and publish Platform Events

  • Debug log retrieval and trace-flag management

  • Access operations (permission sets, permission set groups, user creation)

  • Live Salesforce skill discovery from forcedotcom/sf-skills (including Flow and OmniStudio-focused skills)

  • Live LWC recipe discovery from trailheadapps/lwc-recipes

  • Live Trailhead catalog search and content fetch via public Trailhead MCP mcp.trailhead.salesforce.com

  • Playwright-based UI automation for non-API tasks, including tooling to build and run test automation

  • Support for Git-backed Salesforce projects

Use your MCP client’s tool browser to see the current live tool catalog. Trailhead-prefixed tools (sf_trailhead_*) call the external Trailhead MCP endpoint and return Trailhead server capabilities/content, not the full local Salesforce MCP Pro tool catalog.

Requirements

  • Node.js >=22.19.0

  • Salesforce CLI (sf) installed and authenticated

Quick Start

npm install
npm start

Portable MCP Config (Option A)

This avoids machine-specific absolute paths when repos are cloned with a shared sibling layout.

Assumed layout:

  • .../<salesforce-project>

  • .../SalesforceMCP

From <salesforce-project>/.cursor/mcp.json:

{
  "mcpServers": {
    "Salesforce MCP Pro": {
      "command": "node",
      "args": ["../SalesforceMCP/src/index.js"]
    }
  }
}

For Claude Desktop, use the same command and args under mcpServers.

Safety Model

  • Org allow-listing (ALLOWED_ORGS)

  • Global read-only mode (READ_ONLY)

  • Protected-org mutation confirmation (enabled by default)

Key environment variables:

  • SF_CLI_PATH (optional)

  • ALLOWED_ORGS (optional)

  • READ_ONLY (optional)

  • ENFORCE_PROTECTED_ORG_CONFIRM (optional, default true)

  • PROTECTED_ORG_PATTERNS (optional, default prod,production,live)

  • GITHUB_TOKEN (optional, recommended for higher GitHub API rate limits when using skill-discovery tools)

UI Automation Notes

  • UI tools use Playwright and are designed for deterministic, definition-driven browser automation.

  • ui_session_start defaults to headless: true for normal runs.

  • Use slowMoMs during debugging:

    • 0 for normal speed

    • 100-250 for human-observable debugging

  • For visual troubleshooting, set headless: false and capture screenshots with ui_screenshot.

  • Ensure Playwright browser binaries are installed in the runtime environment used by this server.

Documentation Map

  • Query/search examples: docs/search-examples.md

  • Agentforce examples: docs/agentforce-examples.md

  • Tooling API examples: docs/tooling-api-examples.md

  • DX operations examples: docs/dx-operations-examples.md

  • Chatter examples: docs/chatter-examples.md

  • Packaging examples: docs/packaging-examples.md

  • CRUD examples: docs/crud-examples.md

  • Skills discovery examples: docs/skills-examples.md

  • LWC recipes examples: docs/lwc-recipes-examples.md

  • Multi-Framework examples: docs/multiframework-examples.md

  • Trailhead examples: docs/trailhead-examples.md

  • UI automation examples: docs/ui-examples.md

  • Version control examples: docs/version-control-examples.md

  • Nebula Logger examples: docs/nebula-examples.md

Local Smoke Test

  1. Start the server with npm start.

  2. Confirm stderr shows Salesforce MCP Pro is running on stdio.

  3. In your MCP client, confirm the Salesforce MCP Pro server is connected and tools are available.

Available Tools

72 tools
git_file_historyB

List commit history entries for a specific file path.

Inputs: filePath (required), limit, directory.

git_last_commit_filesB

List file paths changed in the latest commit (names only).

Inputs: directory.

git_last_commit_for_fileB

Find the most recent commit where a specific file changed.

Inputs: filePath (required), directory.

git_recent_commitsB

List recent git commits with author, date, and message.

Inputs: limit, directory.

git_statusC

Get git branch and working tree status in the current project.

Inputs: directory.

sf_agent_activateB

Activate an Agentforce agent by API name and optional version.

Inputs: apiName (required), confirm, version, directory, +2 more.

sf_agent_createB

Create an Agentforce agent from a local agent spec file.

Inputs: name (required), specPath (required), apiName, confirm, +4 more.

sf_agent_deactivateB

Deactivate an Agentforce agent by API name.

Inputs: apiName (required), confirm, directory, targetOrg, +1 more.

sf_agent_preview_endB

End Agentforce preview sessions and return trace info.

Inputs: endAll, apiName, noPrompt, directory, +4 more.

sf_agent_preview_sendB

Send an utterance to an existing Agentforce preview session.

Inputs: utterance (required), apiName, confirm, directory, +4 more.

sf_agent_preview_sessionsB

List locally cached Agentforce preview sessions.

Inputs: directory.

sf_agent_preview_startC

Start a programmatic Agentforce preview session.

Inputs: apiName, confirm, directory, targetOrg, +5 more.

sf_assign_permission_setB

Assign one or more permission sets to a user in Salesforce.

Inputs: permissionSetNames (required), confirm, directory, targetOrg, +1 more.

sf_assign_permset_groupC

Assign one or more permission set groups to a user in Salesforce.

Inputs: permissionSetGroupNames (required), confirm, directory, targetOrg, +1 more.

sf_build_apex_cursor_queryC

Build SOQL and Apex cursor code templates for large-record processing.

Inputs: limit, fields, orderBy, fetchSize, +9 more.

sf_cancel_deployC

Cancel a running deployment by deploy job id.

Inputs: jobId (required), confirm, directory, targetOrg.

sf_chatter_get_record_feedC

Get Chatter feed elements for a Salesforce record.

Inputs: recordId (required), pageSize, directory, targetOrg.

sf_chatter_post_to_groupC

Post a Chatter feed item to a collaboration group.

Inputs: groupId (required), message (required), title, confirm, +2 more.

sf_chatter_post_to_recordB

Post a Chatter feed item to a Salesforce record feed.

Inputs: message (required), recordId (required), confirm, directory, +1 more.

sf_create_recordsC

Create records in a Salesforce object using jsforce.

Inputs: records (required), objectName (required), confirm, directory, +1 more.

sf_create_userC

Create a Salesforce User record with profile and locale settings.

Inputs: alias (required), email (required), lastName (required), username (required), +12 more.

sf_delete_recordsC

Delete records in a Salesforce object using jsforce.

Inputs: ids (required), objectName (required), confirm, directory, +1 more.

sf_deploy_metadataC

Deploy Salesforce metadata from local project files.

Inputs: wait, tests, confirm, manifest, +5 more.

sf_deploy_metadata_validateB

Validate Salesforce metadata deployment without committing changes.

Inputs: wait, tests, confirm, manifest, +5 more.

sf_describe_objectC

Describe a Salesforce object schema including fields and relationships.

Inputs: objectName (required), directory, targetOrg.

sf_disable_debug_logsB

Disable debug logs by TraceFlag id or for all active flags of a user.

Inputs: confirm, username, directory, targetOrg, +1 more.

sf_enable_debug_logsC

Enable debug logs for a user by creating a TraceFlag.

Inputs: username (required), confirm, logType, directory, +3 more.

sf_execute_anonymous_apexC

Execute anonymous Apex in the target org.

Inputs: apexCode (required), confirm, directory, targetOrg.

sf_execute_apex_cursor_queryC

Build Apex cursor query code and execute it via anonymous Apex in one step.

Inputs: limit, fields, confirm, orderBy, +12 more.

sf_execute_batch_jobB

Run an Apex batch class immediately via Database.executeBatch.

Inputs: batchClassName (required), confirm, directory, scopeSize, +1 more.

sf_execute_named_queryC

Execute a Salesforce Named Query API resource by query API name.

Inputs: queryApiName (required), directory, targetOrg, parameters.

sf_generate_flow_definitionC

Generate a baseline Salesforce Flow metadata definition file.

Inputs: flowApiName (required), label, confirm, directory, +9 more.

sf_generate_lwc_componentC

Generate baseline LWC component files in the Salesforce project.

Inputs: componentName (required), confirm, createCss, directory, +8 more.

sf_get_debug_logB

Read the body and metadata for a specific Apex debug log id.

Inputs: logId (required), directory, targetOrg.

sf_get_deploy_statusC

Get deployment status and details by deploy job id.

Inputs: jobId (required), wait, directory, targetOrg.

sf_get_usernameB

Resolve username and alias for a target org or default target org.

Inputs: directory, targetOrg, defaultTargetOrg.

sf_get_user_object_create_accessC

Check object-level create permission for a single user across one or more objects.

Inputs: userId (required), objectApiNames (required), directory, targetOrg.

sf_get_user_record_accessB

Check read/edit/delete/transfer access for a single user across up to 200 records.

Inputs: userId (required), recordIds (required), directory, targetOrg.

sf_is_field_nillableB

Determine whether a Salesforce field is nillable on a specific object.

Inputs: fieldName (required), objectName (required), directory, targetOrg.

sf_list_all_orgsA

List authenticated Salesforce orgs from Salesforce CLI.

Inputs: no named parameters.

sf_list_debug_logsB

List recent Apex debug logs from the target org.

Inputs: limit, username, directory, targetOrg.

sf_list_metadataC

List metadata components by metadata type for discovery.

Inputs: metadataType (required), folder, directory, targetOrg, +1 more.

sf_package_createB

Create a second-generation Salesforce package in a Dev Hub org.

Inputs: name (required), path (required), packageType (required), confirm, +7 more.

sf_package_installC

Install a package version in a target org.

Inputs: package (required), wait, confirm, noPrompt, +7 more.

sf_package_listC

List second-generation packages available in a Dev Hub org.

Inputs: verbose, directory, targetDevHub.

sf_query_orgC

Run a SOQL query against a Salesforce org.

Inputs: query (required), directory, targetOrg.

sf_query_org_paginatedB

Run paginated SOQL queries with explicit nextRecordsUrl support.

Inputs: query (required), pageSize, directory, targetOrg, +2 more.

sf_remove_permission_setB

Remove one or more permission sets from a user in Salesforce.

Inputs: permissionSetNames (required), confirm, directory, targetOrg, +1 more.

sf_remove_permset_groupC

Remove one or more permission set groups from a user in Salesforce.

Inputs: permissionSetGroupNames (required), confirm, directory, targetOrg, +1 more.

sf_retrieve_metadataC

Retrieve metadata from a Salesforce org to local project files.

Inputs: wait, manifest, metadata, directory, +3 more.

sf_run_apex_testC

Run Apex tests in a Salesforce org.

Inputs: wait, tests, suites, directory, +3 more.

sf_run_apex_test_suiteC

Run Apex test suites (suite-first input) in a Salesforce org.

Inputs: suites (required), wait, directory, targetOrg, +1 more.

sf_schedule_apex_jobC

Schedule a Schedulable Apex class using a cron expression.

Inputs: jobName (required), className (required), cronExpression (required), confirm, +2 more.

sf_search_orgC

Run a SOSL search against a Salesforce org.

Inputs: search (required), directory, targetOrg.

sf_skills_getB

Fetch a Salesforce skill definition (SKILL.md) from forcedotcom/sf-skills.

Inputs: skillName (required), includeReferences.

sf_skills_list_domainsC

List skill domains and sample skills from forcedotcom/sf-skills.

Inputs: limitPerDomain.

sf_tooling_describe_objectB

Describe a Tooling API object schema by object name.

Inputs: objectName (required), directory, targetOrg.

sf_tooling_get_apex_classB

Fetch ApexClass metadata and body by class name via Tooling API.

Inputs: className (required), directory, targetOrg.

sf_tooling_list_objectsB

List Tooling API sObject types available in the target org.

Inputs: directory, targetOrg.

sf_tooling_queryC

Run a SOQL query against Salesforce Tooling API objects.

Inputs: query (required), directory, targetOrg.

sf_unschedule_apex_jobC

Unschedule one or more scheduled jobs by jobId or jobName.

Inputs: jobId, confirm, jobName, directory, +2 more.

sf_update_recordsC

Update records in a Salesforce object using jsforce.

Inputs: records (required), objectName (required), confirm, directory, +1 more.

sf_upsert_recordsC

Upsert records in a Salesforce object using jsforce.

Inputs: records (required), objectName (required), externalIdField (required), confirm, +2 more.

ui_actionD

Perform a direct UI action on a selector.

Inputs: action (required), selector (required), value, sessionId, +1 more.

ui_assertC

Assert UI conditions (visibility and text).

Inputs: selector (required), visible, sessionId, timeoutMs, +1 more.

ui_extractC

Extract text or attributes from page elements.

Inputs: selector (required), all, attribute, sessionId, +1 more.

ui_navigateB

Navigate the current UI session to a URL or path.

Inputs: url, path, sessionId, timeoutMs, +1 more.

ui_run_flowC

Run a named UI flow from ui-definitions/flows.json.

Inputs: flowName (required), confirm, sessionId, variables.

ui_screenshotC

Capture a screenshot from the current UI session.

Inputs: fileName, fullPage, sessionId.

ui_session_startC

Start a Playwright browser session for UI automation.

Inputs: baseUrl, headless, slowMoMs, viewportWidth, +1 more.

ui_session_stopB

Stop and dispose a Playwright UI session.

Inputs: sessionId.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    C2.7/5.0

    Scored across 72 tools

    Disambiguation4/5

    Most tools have clear, distinct resource-action pairs, but some overlap exists: sf_execute_apex_cursor_query and sf_build_apex_cursor_query both handle Apex cursor logic, and ui_action is somewhat vague. Despite the large count, descriptions generally disambiguate well.

    Naming Consistency4/5

    The sf_ prefix with verb_noun pattern is consistent (sf_list_orgs, sf_create_records), and ui_/git_ prefixes also follow a similar pattern. Minor inconsistency like sf_agent_preview_sessions (noun instead of verb) and sf_tooling_list_objects vs sf_list_metadata slightly deviate but overall naming is predictable.

    Tool Count2/5

    72 tools is excessive, far beyond the typical well-scoped range. While the server aims to be comprehensive for Salesforce, the sheer number creates cognitive overload and makes tool selection harder, even if each tool individually serves a purpose.

    Completeness4/5

    The tool surface is extensive, covering queries, CRUD, metadata deploy/retrieve, Apex execution, agents, tests, permission management, UI automation, and git workflows. Minor gaps exist (e.g., no list permission sets tool) but most core Salesforce development workflows are supported.

    Maintenance

    ActivitySlowing
    ResponsivenessLimited data

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      B
      quality
      C
      maintenance
      A Model Context Protocol server that enables LLMs to interact with Salesforce data through SOQL queries, SOSL searches, and various API operations including record management.
      14
      197 PyPI
      179
      MIT
    • F
      license
      Not graded
      quality
      D
      maintenance
      A Model Context Protocol server that enables interaction with Salesforce through REST API using jsforce, allowing users to execute SOQL queries, retrieve object metadata, and perform CRUD operations on Salesforce records.
      -
    • A
      license
      Not graded
      quality
      D
      maintenance
      A TypeScript implementation of a Model Context Protocol server that enables LLMs to interact with Salesforce data through SOQL queries, SOSL searches, and CRUD operations.
      2
      MIT