Build visual content that stays editable
Create images and videos in Studio, automate typed templates through the V2 API, or let an Agent work through MCP.
JsonCut uses one HTML-first project format across Studio, templates, the V2 API and MCP. An image or video created by an integration can still be opened and edited in Studio.
Choose the right workflow
| Goal | Start here | Stored? | Best for |
|---|---|---|---|
| Create and keep editing | Project quickstart | Yes | Studio, applications and iterative Agents |
| Generate repeatable variants | Typed templates | Template and render history | Personalization, CSV and bulk automation |
| Produce one disposable result | Inline rendering | Temporary by default | Small server-side one-shots |
| Let an Agent design and review | MCP setup | Projects by default | ChatGPT, Claude and other MCP clients |
The four-step API loop
- Upload reusable assets with
POST /api/v2/media. - Create a project, publish a typed template, or submit a bounded inline source.
- Queue a render. Invalid source, unavailable media and incompatible output settings are rejected before work begins.
- Poll the render resource or receive a signed webhook, then download the artifact.
Every developer request uses X-API-Key. Keep the key on your server. Use an Idempotency-Key for render creation and obey the returned retryable flag when handling errors.
Learn the project format
- Create with V2: the complete example-driven authoring path.
- Project source and canvas: the outer
jsoncut-htmlcontract. - Layers and media: text, shapes, SVG, images, video, audio and HTML.
- Motion and transitions: keyframes, easing and clip-to-clip transitions.
- Variables and templates: typed inputs for repeatable output.
Contracts and compatibility
These docs cover only the supported public developer surface. Product administration and Studio-only workflows are documented inside the product where they are used.