Goals:

• Enable cast actions to present a frame for multi-step user interactions

Some actions may require user input or other multi-step interactions. For example:

• Tipping a specific amount
• Saving a bookmark to a folder
• Minting a cast image as an NFT

In general, it’s a good principle for actions that have side effects to present some form of confirmation UI.

Cast actions that require user input or confirmation should use a frame to collect it. Since the frame is not a “real” cast embed, clients will have to render the frame in a special context, like a bottom sheet in a mobile app or a modal on web.

Cast action frames rendered in a special context outside of a cast embed.

Specification

Action servers may return a response in the following form to display a frame:

{"type": "frame", "frameUrl": "<https://frame.example.com>"}

Client applications must:

• Validate the response:
  • Type 'frame'
    • must include frameUrl
    • frameUrl must begin with https://
  • Type 'message'
    • must include a message string < 80 characters