JsonCut
Agents · 9 min

Create images and video through MCP

Connect an agent, keep long work project-based and review visible results before rendering.

MCP gives a compatible agent a focused visual tool surface. Short work can be created in one pass, while longer videos remain manageable through scenes, patches and review frames.

Connect the V2 MCP endpoint

Add the JsonCut MCP endpoint and authenticate with an API key from Studio. The V2 server exposes projects, media, templates, visual review and rendering without internal Studio tools.

Use projects for iterative work

For a short graphic, one creation call may be enough. For a longer video, create the project first, outline scenes and patch only the scene or group currently being refined.

Give the agent visual evidence

Render an image snapshot or a representative video frame grid. Ask the agent to check alignment, safe areas, media crops and timing before it claims the work is finished.

Continue with the real interface.

The V2 documentation turns every public endpoint into an interactive reference.