Workflow
Last updated on
Jul 28, 2026
Discover how the Run-Workflow MCP connects AI agents to Adobe’s creative workflows and automates complex tasks through natural language.
Run-Workflow MCP connects Model Context Protocol (MCP)-compatible AI agents and Adobe’s creative workflow services. It allows AI assistants like Claude Code, Claude Desktop, and Cursor to discover available workflow capabilities from Firefly Creative Production, select appropriate actions, and execute workflows on your behalf using natural language instructions.
With the connector, you can describe what you want to accomplish in everyday language while your AI agent handles the technical implementation.
How Run-Workflow MCP works
Run-Workflow MCP provides an architecture where your AI agent communicates with a hosted MCP server that exposes Adobe workflow capabilities as discoverable tools.
Based on your request, the AI agent analyzes your intent and determines the required workflow actions. The server then handles workflow submission, monitors execution progress asynchronously, and returns results when processing completes. This asynchronous model supports long-running operations like video processing or large batch jobs without waiting for a response.
Run-Workflow lifecycle
A typical workflow progresses through these stages:
- Provide workflow inputs: Share your workflow request, the required files, data, or URLs with the agent.
- Create or select a workflow: Depending on your request, your AI assistant either creates a new workflow or selects a prebuilt workflow designed for common creative production tasks.
- Submit the workflow: Processing begins in the background and returns a tracking identifier immediately.
- Monitor progress: Your AI agent checks the status of your ask until it completes.
- Retrieve outputs: You can download finished output files through temporary links.
- Reuse the workflow (optional): Publish completed workflows to obtain reusable workflow identifiers for future use.
For example, if you ask your agent to remove the background from an image and place it on a white backdrop, the agent executes the workflow and shares the completed result.
Supported MCP clients
Run-Workflow MCP supports the following MCP-compatible AI clients:
- Claude Code
- Claude Desktop
- Cursor
Each client connects to the same hosted Run-Workflow MCP service. Installation steps differ between clients, but available workflow capabilities remain consistent across all platforms. Learn how to install and get started with the Run-Workflow MCP.
What you can do with Run-Workflow MCP
Run-Workflow MCP supports the following creative workflows:
|
|
Example prompt |
|
Remove backgrounds from images |
Remove the background from this shoe photo and place it on a white backdrop. |
|
Generate or modify image content |
Generate a lifestyle background for this backpack and place the product on a hiking trail. |
|
Composite images into templates |
Insert this product image into my retail flyer template and preserve the existing text layout. |
|
Process video workflows |
Create 16:9, 9:16, and 1:1 versions of this promotional video. |
|
Chain multiple creative actions into a single workflow |
Remove the background, generate a holiday-themed scene, enhance the lighting, and create both web and print versions. |
|
Save completed workflows for future reuse |
Publish this workflow so my team can reuse it for future product launches. |
|
Search and use featured workflow templates |
Find a featured workflow for product retargeting ads and run it using these product images. |
You don't need to know tool names — just describe your goal. Your AI agent automatically discovers available workflow actions and determines which tools to use based on your request.
Turn text into art with Adobe Firefly
Create beautiful images and video clips from text prompts using generative AI.
