canva

Automate Canva design exports by using a typed Python SDK that calls MCP tools directly, eliminating model round-trips per tool invocation. This approach reduces latency and token usage while handling uploads, searches, and folder management programmatically.

Yes, you can use Python async methods to search and merge Canva designs programmatically. The SDK exposes Canva MCP tools as a typed asynchronous Python class compatible with the fastmcp client for upload, search, get, merge, export, and folder management tasks.

The Canva MCP SDK handles OAuth authentication with token persistence to a local directory and manages pagination continuation tokens automatically. It enforces safety rules like distinguishing template versus design searches and requires explicit confirmation for destructive operations like page deletion.

The best way to import assets and upload images to Canva from a public URL is instantiating a CanvaApp with default OAuth and calling the upload_asset_from_url method. This typed async function handles the HTTPS image upload directly within your agent workflow code.

The SDK handles JSON parsing errors by attempting to parse tool outputs as JSON and falling back to structured text results for robust downstream handling. This error tolerance ensures your automation workflows continue functioning even when Canva tool outputs are not valid JSON.

You need the fastmcp client because the typed asynchronous methods in the SDK are specifically designed to be compatible with it. This compatibility allows your agent workflows to call Canva MCP tools directly in Python code without round-tripping to a model for each invocation.