Does Penpot MCP support generating designs from a prompt?

No. Penpot MCP does not have a generation tool. Its five tools, high_level_overview, penpot_api_info, execute_code, export_shape, and import_image, focus on reading, modifying, and exporting an existing file that is open in the Penpot editor. Canva MCP provides generate-design and generate-design-structured for prompt-based creation.

Does Canva MCP give access to design file internals the way Penpot's execute_code does?

Partially. get-design-content reads elements and content, get-design-pages retrieves page information, and get-design-thumbnail returns a preview. But Canva does not expose an arbitrary code execution tool; edits go through the start-editing-transaction and perform-editing-operations and commit-editing-transaction flow, which is more controlled than Penpot's scriptable plugin access.

Penpot vs Canva

Penpot MCP and Canva MCP are both official design servers, but they take fundamentally different approaches to what an agent can do. Penpot's server exposes five tools that give an agent deep programmatic access to a live design file: high_level_overview orients the agent to the file's structure, penpot_api_info surfaces the plugin API, execute_code runs JavaScript against the file through the Penpot plugin (enabling queries, edits, and element creation), export_shape renders a shape to an image, and import_image brings assets into the design. Canva's server is broader in workflow scope with over 30 tools: generate-design and generate-design-structured create new designs from prompts, autofill-design populates brand templates, start-editing-transaction and perform-editing-operations and commit-editing-transaction handle transactional edits, search-designs and get-design-content read existing work, export-design outputs to PDF or image, and additional tools manage assets, brand kits, folders, comments, and resizing. Penpot is open-source and file-centric; Canva is a SaaS platform with a richer workflow surface.

How they compare

Dimension	Penpot	Canva
Platform	Open-source Penpot, self-hostable or at design.penpot.app. The server connects to a live file through the Penpot plugin over WebSocket.	Canva SaaS at canva.com. The server is a hosted remote endpoint at https://mcp.canva.com/mcp, authorized through OAuth with Dynamic Client Registration.
Design generation	No prompt-to-design generation. An agent reads and modifies existing files through execute_code and can export shapes with export_shape.	generate-design creates a design from a prompt, generate-design-structured follows a specified structure, and create-design-from-candidate builds from a generated candidate or template.
Editing model	execute_code runs arbitrary JavaScript through the Penpot plugin, giving the agent full programmatic access to query, transform, and create design elements in the open file.	Transactional editing: start-editing-transaction, then perform-editing-operations in a batch, then commit-editing-transaction to save or cancel-editing-transaction to discard.
Asset and brand management	import_image brings an asset into the design (local file import in local mode). No brand kit or template catalog tools.	upload-asset-from-url, get-assets, list-brand-kits, search-brand-templates, get-brand-template-dataset, and autofill-design let the agent work with the full Canva brand system.
Export and review	export_shape renders and exports a specific shape as an image. No comment or review tools.	export-design outputs a full design to PDF or image formats; comment-on-design, reply-to-comment, and list-comments support a review workflow.

Verdict

Choose Penpot MCP when the agent's job is to understand and modify an existing open-source design file programmatically: high_level_overview and execute_code give it deep, scriptable access to layouts, components, and styles in the live file, with export_shape to render results. Choose Canva MCP when the agent needs to generate new designs from prompts, autofill brand templates, manage assets and brand kits, run transactional edits, handle review comments, or export finished work, all against a Canva account without installing anything locally. Penpot suits design-to-code and file-manipulation workflows on an open platform; Canva suits asset creation and brand-workflow automation on a SaaS platform.

FAQ

Does Penpot MCP support generating designs from a prompt?: No. Penpot MCP does not have a generation tool. Its five tools, high_level_overview, penpot_api_info, execute_code, export_shape, and import_image, focus on reading, modifying, and exporting an existing file that is open in the Penpot editor. Canva MCP provides generate-design and generate-design-structured for prompt-based creation.
Does Canva MCP give access to design file internals the way Penpot's execute_code does?: Partially. get-design-content reads elements and content, get-design-pages retrieves page information, and get-design-thumbnail returns a preview. But Canva does not expose an arbitrary code execution tool; edits go through the start-editing-transaction and perform-editing-operations and commit-editing-transaction flow, which is more controlled than Penpot's scriptable plugin access.