The Sales Layer MCP Server connects your Sales Layer catalog and DAM libraries to AI assistants and developer tools that speak the Model Context Protocol (MCP).
It exposes Catalog and DAM capabilities as MCP tools, so an assistant can answer questions about products, categories, variants, and custom entities; inspect layouts and changelogs; manage images and files; and — when explicitly authorized — modify Sales Layer data.
The MCP Server is a client-facing access layer. The Catalog REST API and DAM REST API remain the sources of truth for endpoints, parameters, schemas, and error behavior.
| Layer | What it is | When you use it |
|---|---|---|
| Catalog REST API | HTTPS endpoints authenticated with X-API-KEY. | You are building an integration that calls fixed endpoints from your own code. |
| DAM REST API | HTTPS endpoints for image and non-image file libraries. | Your integration manages digital assets directly. |
| MCP Server | MCP tools consumed by an AI client, authenticated with OAuth 2.0. | You want an assistant to work with catalog data and DAM assets conversationally. |
MCP tools are not additional REST endpoints. Each tool is backed by Catalog or DAM behavior, so the constraints documented in the REST guides — filtering, field selection, pagination, and rate limiting — apply to what a tool can return.
If a workflow calls fixed endpoints and does not need an assistant to decide what to do, use the REST API directly. It is simpler and does not require an interactive authorization step.
Read-oriented work is the common starting point:
- Ask questions about the catalog in natural language, including across multiple languages.
- Run data quality checks, such as finding products with missing attributes or untranslated fields.
- Explore catalog structure — available tables, fields, field types, and configured languages.
- Inspect Attribute Sets, form layouts, and resource changelogs.
- Find DAM images and files, check their status, and identify unlinked assets.
- Generate summaries and reports over products, categories, and variants.
Write-oriented work requires the Full access profile and explicit organizational approval:
- Create, update, or delete catalog records, layouts, images, and files.
- Support AI-assisted enrichment workflows with a human confirmation step.
See Access profiles and permissions for exactly which tools each profile allows.
| Client | Connection |
|---|---|
| Claude Web | Remote |
| Claude Desktop | Local package or extension |
| ChatGPT | Remote |
| Microsoft Copilot Studio | Remote |
| Google Gemini | Remote |
| Cursor | Remote |
| VS Code and GitHub Copilot | Remote |
| n8n | Remote, when the platform supports MCP client nodes with OAuth 2.0 |
| Other MCP-compatible clients | Remote or local |
Any client that supports remote MCP servers with OAuth 2.0 and PKCE can connect. Clients without PKCE and S256 support cannot use the remote server.
The endpoint differs per client. See Quickstart for the correct endpoint for yours.
Two profiles control what the assistant is allowed to do:
- Read-only — query and analyze catalog data. No modifications reach your PIM. Recommended for a first connection.
- Full access — read plus the write tools. Use it only once your team has approved the workflow.
- Request a Catalog Token. See Authentication.
- Connect your client. See Quickstart.
- Start with read-only access and confirm the assistant can list the available tools.
Client-by-client setup instructions, account configuration, and support escalation are maintained in the Sales Layer Support Center.