Should I use Super Agent, Flow, Nodes, API, or MCP?
Choose the surface based on how defined and repeatable your work is.
Your goal | Best place to start |
|---|---|
Complete a one-time or open-ended task | Super Agent |
Repeat the same defined process | Flow |
Run one model or utility directly | Nodes |
Call Magica from your own software | API |
Let an MCP-compatible AI client use Magica | MCP |
Use Super Agent for outcomes
Choose Super Agent when you know the result you want but do not want to design every step. It can research, browse, work with files, create content, build code, use connected apps, and coordinate multiple tools.
Example: “Research five competitors, compare their positioning, and create an executive summary.”
Use Flow for repeatable processes
Choose Flow when the steps should run the same way each time. You connect models and utilities on a visual canvas, configure the inputs, and run the workflow again with new data.
Example: “For each product description, generate ad copy, an image, and a short video.”
Use Nodes for one direct model run
Choose Nodes when you already know the type of model or tool you need and do not need an agent or multi-step workflow.
Example: run an image generator, transcription model, LLM, or media utility directly.
Use API or MCP for external access
Use the API when your application or backend should call Magica programmatically. Use MCP when a compatible AI client should discover and use Magica tools and workflows.
The public API and MCP surfaces focus on models, Nodes, and Flow workflows. Super Agent tasks are not exposed as a public API.
View API and MCP documentation
Updated on: 23/07/2026
Thank you!