Natural Language Prompts

Prompt mode is for fast exploration. Pass a prompt string and ScreenFramed resolves it into normal capture parameters before validation and rendering.

json
{  "prompt": "Capture the ScreenFramed gallery as a polished browser hero with a dark glass background",  "url": "https://screenframed.com/gallery"}

How it resolves

Prompt mode is a convenience layer over the same public API. The resolved params are merged with your explicit params, and explicit params win.

flowchart LR
A[prompt] --> B[Resolve params]
B --> C[Merge explicit fields]
C --> D[Validate]
D --> E[Render]

Good prompts

Examples:

json
{  "url": "https://example.com",  "prompt": "Make a 16:9 product hero with browser chrome, dark premium background, and a soft shadow",  "aspect_ratio": "16:9"}

json
{  "url": "https://example.com/pricing",  "prompt": "Capture only the pricing table as a clean documentation image",  "selector": "[data-capture='pricing-table']"}

json
{  "url": "https://example.com/mobile",  "prompt": "Create an app store phone screenshot with a short headline at the top",  "device": "iphone-16-pro",  "aspect_ratio": "9:16"}

Debug resolved params

Use show_resolved_params when developing prompt workflows:

json
{  "url": "https://example.com",  "prompt": "Make a cinematic product hero",  "show_resolved_params": true}

The response includes resolved_params, which you can copy into a stable production request.

Use prompt mode to discover a direction, then save the resolved explicit params for repeatable rendering in CI or product code.