The Vantage View | Salesforce

Salesforce Headless 360 for Developers: MCP Servers and Data 360 APIs

Written by David Cockrum | Sep 6, 2026, 11:59:59 AM

This is Part 2 of Vantage Point's Headless 360 series. Part 1, "Salesforce Headless 360: Architecture and Readiness Guide," explains what the architecture is, what is generally available, and how to govern it. This post is the builder's companion: the hands-on resources, API surfaces, and workshops that take you from reading about Headless 360 to shipping with it. Later parts cover Headless 360 for admins, HXL and Mosaic, headless commerce, and headless identity.

With Dreamforce weeks away, Salesforce's developer story has shifted fast. Headless 360 makes every major Salesforce capability available as an API, a Model Context Protocol (MCP) tool, or a CLI command — so your code, and your AI coding assistants, can reach the platform without a browser in the loop. For developers, architects, and technical leads, the question is no longer whether to engage with this model — it is where to start. This guide tours what Salesforce has shipped for builders: the Headless 360 Developer Center, hosted MCP servers, the hands-on workshop, Data 360 APIs, and AI-assisted development in practice.

Quick Answer

Salesforce Headless 360 for developers is the platform's full capability surface — more than 60 MCP tools, 30+ preconfigured coding skills, 4,000+ existing APIs, and 220+ CLI commands — exposed to any authenticated caller your trust layer authorizes, including AI coding assistants like Claude, ChatGPT, and Cursor. The practical starting point is the Headless 360 Developer Center, followed by the hosted-MCP workshop and the Data 360 developer guide.

TL;DR

  • What it is: Headless 360 exposes Salesforce data, logic, and experiences as APIs, MCP tools, and CLI commands — no browser required.
  • Why it matters: AI coding agents can build against your org, and business agents can serve users from it on surfaces like Slack, ChatGPT, and mobile.
  • Where to start: The Headless 360 Developer Center, the "Explore Salesforce Hosted MCPs" workshop, and the Data 360 development guide.
  • Decision point: Confirm which components are GA for your edition — Part 1 covers availability and governance.
  • How Vantage Point helps: Senior consultants design the architecture, identity model, and pilot plan so your team builds on solid ground.

What Did Salesforce Ship for Developers With Headless 360?

Salesforce's developer blog puts it plainly: Headless 360 makes every major capability "openly composable by agents, not just by code paths a human developer wrote in advance." In What Salesforce Headless 360 Means for Developers, the Developer Advocacy team frames the shift around two kinds of MCP tools, distinguished by who is calling and why:

MCP lane Who calls What it does Example
Build-time (coding agents) Your AI coding assistant Reaches into your org to build, test, and deploy An assistant drafts an Apex class, runs SOQL, and deploys metadata through MCP tools
Runtime (business agents) An agent serving your users Reads permitted data and invokes approved operations A support agent pulls a case summary into a Slack thread

The build-time surface is mature today; the runtime surface already supports straightforward use cases, with a broader vision of delivering the same capability on any surface. And while agents now produce first drafts of Apex classes, LWCs, and test suites, your platform expertise does not become obsolete — it becomes the review layer that makes agent output safe to ship.

What Is the Headless 360 Developer Center?

The Headless 360 Developer Center is Salesforce's centralized destination for all of this content. If you bookmark one page before Dreamforce, make it this one:

  • Get Started: the Get Started with Headless 360 guide, a free trial, hands-on workshops, the Hosted MCP Servers reference, and release notes.
  • MCPs and Skills: the Salesforce DX MCP Server (gives AI coding assistants access to your orgs — deploy metadata, run SOQL and Apex, build LWCs), Agentforce Vibes MCP servers, MuleSoft and Heroku MCP servers, and the Data 360 MCP Server for real-time customer context.
  • Development Tools: Agentforce Vibes (describe what you need; Agentforce builds, tests, and deploys the code), the full API catalog, the Salesforce CLI quick start, multi-framework React development, the HXL Playground, and the next-generation DevOps Center.
  • Trailhead and video: the Introduction to Salesforce Headless 360 module, Agentforce Vibes modules, and the Headless 360 Decoded video series.

The center also links the Headless 360 MCP Server beta announcement and the newsroom announcement. Treat it as your adoption guide as new tools ship.

How Do Salesforce Hosted MCP Servers Work?

Hosted MCP servers are the inbound path: an external MCP-compatible client discovers and calls approved Salesforce capabilities over a standardized protocol, with no custom integration code written in advance for each one. As covered in Part 1 of this series, Salesforce announced general availability of Hosted MCP Servers for Enterprise Edition orgs and above on April 29, 2026 — but individual servers, tools, and metadata features carry their own statuses, so verify each one in the Hosted MCP Servers reference before you build.

Three things matter for developers:

  1. Authentication and authorization still apply. Hosted MCP requests run as an authenticated identity, and object- and field-level security still shape what comes back. Design the identity model first — named user context where accountability matters, or a narrowly scoped service identity where it does not.
  2. You can build custom servers. The workshop below walks through creating a custom Flow-backed MCP server — how you expose your own business logic as agent-callable tools.
  3. MCP is not the answer to every integration. For deterministic system-to-system workloads, traditional APIs or MuleSoft may still fit better. Our post on Salesforce MCP vs. MuleSoft breaks down that decision, and Wiring Salesforce to Claude via MCP: Field Lessons covers practical connection patterns.

How Do You Get Hands-On With the Hosted-MCP Workshop?

The fastest way to build real intuition is the Headless 360 Workshop, starting with the Explore Salesforce Hosted MCPs chapter. In one sitting you will:

  1. Configure Salesforce MCP servers — enable and connect hosted MCP servers in a dedicated environment.
  2. Test the servers against live data — explore real Sales Cloud data in the Headless 360 Playground through an MCP client.
  3. Create a custom Flow-backed MCP server — expose your own automation as a tool an agent can discover and call.

The broader workshop path continues into Agentforce Vibes and React, from connection through deployment of an AI-assisted Salesforce experience. A shorter Try Headless 360 workshop focuses on securely connecting Claude to Salesforce through hosted MCPs. Block two hours before Dreamforce; the hands-on context will change how you evaluate the announcements.

What Can You Build With Data 360 APIs?

Data 360 (formerly Data Cloud) is the context layer of Headless 360, and its developer surface is deep. The Data 360 Development Overview maps the integration points: data lands as data lake objects (DLOs), is mapped to data model objects (DMOs), and is unified through identity resolution — and every step is reachable by API.

The API surfaces to know:

  • Data 360 Connect REST API — the most commonly used API, part of the Salesforce Platform, with REST resources for queries, profiles, calculated insights, identity resolution rulesets, and metadata, plus Apex classes through the ConnectApi namespace. See the Data 360 Connect REST API guide.
  • Data 360 API — tenant-specific endpoints inside the Data 360 tenant for queries, profiles, and metadata at scale.
  • Ingestion APIs — bulk, streaming, and real-time ingestion, authenticated through a connected app with an OAuth token exchange.
  • Data 360 MCP Server — gives agents real-time customer context without bespoke query plumbing.

Salesforce's developer team maintains a Postman collection for the Data 360 APIs, which makes first calls much faster. If your use case involves unified customer data — and most do — data readiness in this layer determines whether the experience built on top succeeds.

How Does AI-Assisted Development Work in Practice?

This is the piece that feels most different day to day. With the Salesforce DX MCP Server connected, an AI coding assistant — Claude, ChatGPT, Cursor, or another MCP-compatible client — can inspect your org's metadata, run SOQL and Apex, and scaffold code against your actual schema rather than a guess. Your job shifts from writing every line to specifying intent, evaluating output, and deciding what fits your system design.

Two adjacent capabilities round out the picture:

  • Agentforce Vibes is Salesforce's own AI-assisted development experience in VS Code: describe what you need, and Agentforce builds, tests, and deploys the code, extensible with MCP servers for additional tools and context.
  • Multi-framework support lets React developers build governed, data-connected applications on Salesforce using patterns they already know — no LWC requirement. That materially changes hiring and staffing constraints for Salesforce projects.

The governance posture from Part 1 applies doubly here: openness is the default, and governance — least privilege, human review gates, sandbox testing — is what makes it safe.

What Are Teams Already Building on Headless 360?

Two public Salesforce stories show the pattern in production. In Salesforce on Salesforce, Salesforce's own IT team decoupled backend transaction logic from the standard UI and now runs employee-facing IT operations natively inside Slack, pulling live data from its Agentforce service instance. And Indeed's engineering team builds, tests, and deploys production Agentforce agents without opening a browser. The common thread: neither team started with a big-bang rebuild — both picked a bounded workflow, exposed it through governed interfaces, and expanded from there.

What Should Developers Do Before Dreamforce?

  1. Bookmark the Headless 360 Developer Center and skim the Get Started guide.
  2. Complete the hosted-MCP workshop — configure servers, test against live data, build one custom Flow-backed server.
  3. Connect one AI coding assistant to a sandbox via the DX MCP Server and run a real task end to end.
  4. Prototype one Data 360 API call with the Postman collection if your roadmap touches unified data.
  5. Review availability and governance in Part 1 of this series so your pilot matches what is actually GA for your edition.
  6. Bring questions to Dreamforce — the gap between the demo and your org is where real planning happens.

How Vantage Point Helps

Vantage Point helps development teams turn Headless 360's building blocks into a governed architecture: selecting the right MCP servers and API surfaces, designing identity and least-privilege access, and structuring sandbox pilots connected to your broader Salesforce implementation and advisory and system integration and data migration strategy. Senior consultants only — no junior handoffs; the experts you meet are the experts who deliver.

FAQ

What is the Salesforce Headless 360 Developer Center?

It is Salesforce's centralized hub for Headless 360 builder content: getting-started guides, hosted MCP server references, workshops, the DX and Data 360 MCP servers, Agentforce Vibes, multi-framework React tooling, Trailhead modules, and release notes.

What MCP servers are available for Salesforce developers?

The main ones are the Salesforce DX MCP Server (org access for AI coding assistants — deploy metadata, run SOQL and Apex, build LWCs), the Data 360 MCP Server (real-time customer context for agents), Agentforce Vibes MCP servers, and MuleSoft and Heroku MCP servers. You can also create custom Flow-backed MCP servers to expose your own business logic as agent-callable tools.

Are Salesforce Hosted MCP Servers generally available?

Yes — Salesforce announced GA for Enterprise Edition orgs and above on April 29, 2026. Individual servers, tools, and metadata features have their own statuses and limits, so confirm each component in the Hosted MCP Servers reference first.

How do I connect Claude or another AI assistant to Salesforce?

Use a hosted MCP server or the DX MCP Server: enable the server in your org, authenticate with an appropriately scoped identity, and add the server to your MCP-compatible client. Salesforce's "Try Headless 360" workshop walks through connecting Claude step by step.

What is the difference between the Data 360 Connect REST API and the Data 360 API?

The Connect REST API runs on the Salesforce Platform and is the most commonly used option, with REST resources and Apex classes for queries, profiles, insights, and metadata. The Data 360 API runs against tenant-specific endpoints inside the Data 360 tenant. Both are documented in the Data 360 developer guide.

Does Headless 360 mean I no longer need to learn LWC or Apex?

No. AI coding agents produce first drafts, but developers specify intent, review output, and own system design — that requires platform expertise. Multi-framework support does mean React skills now transfer directly to Salesforce, so LWC is no longer a hard hiring constraint.

What should I build first with Headless 360?

Start in a sandbox with the hosted-MCP workshop: connect a server, query live Sales Cloud data through the playground, and expose one Flow as a custom MCP server. Then pick one bounded, low-risk workflow and pilot it with least-privilege access before expanding.

Ready to build on Salesforce Headless 360? Whether you are prototyping MCP integrations, evaluating Agentforce Vibes, or designing a composable architecture ahead of Dreamforce, Vantage Point's senior consultants can help you move from workshop to production. Contact Vantage Point to schedule a Headless 360 readiness session, or explore our Salesforce services to see how we support development teams end to end.

Vantage Point is a boutique CRM consulting firm helping businesses transform with Salesforce, HubSpot, and AI — 150+ clients, 400+ engagements, and a 4.71/5 average engagement rating. Learn more at vantagepoint.io.