Hire Team

INDUSAGI DOCUMENTATION

TUI Images

Preserved documentation snapshot. APIs and capabilities differ by edition and version. Check the selected package before implementation. View pinned source.

View TypeScript Framework source on GitHub
On this page

This rebuild does not render images as pixels. There is no Kitty or iTerm2 image protocol, no capability detection, and no image dimension math. Image content is shown as a text placeholder, gated by a showImages flag.

How images are handled

Image content parts (ImageContent from indusagi/ai) are rendered as the literal text [image: <mimeType>] when showImages is true, and omitted entirely when it is false. There is no separate image component — every message and tool-output path runs image parts through the same text-extraction helpers.

The showImages flag

showImages: boolean threads from MessageList down through MessageRow, AssistantMessageView, UserMessageView, CustomMessageView, and ToolResultBlock into the formatting helpers. It controls whether image parts contribute a placeholder line or are dropped.

Placeholder formatting

File: src/react-ink/utils/message-groups.ts

The helpers that turn message content into displayable text emit the placeholder:

  • formatUserContent(content, showImages) — for an ImageContent part, pushes `[image: ${part.mimeType}]` only when showImages is set.
  • formatCustomContent(content, showImages) — same placeholder for custom messages.
  • extractToolText(value, showImages) / extractStructuredText(...) — when walking tool output, an image-typed record yields `[image: ${record.mimeType}]` when showImages is set, and [] otherwise.

File: src/react-ink/utils/tool-display.ts

  • containsImage(value) — detects whether a tool output contains any type: "image" part. The read tool descriptor uses it: when the read result is an image, the card summary becomes the first meaningful output line (or "Read image") instead of a line count, and the raw body is shown unhighlighted.

What is NOT here

The old docs described a src/tui/terminal-image.ts module with Kitty/iTerm2 encoders, capability detection, and pixel-dimension helpers. None of that exists in the current source. Specifically, there is no:

  • detectCapabilities() / getCapabilities() / resetCapabilitiesCache().
  • encodeKitty(...) / encodeITerm2(...) / renderImage(...).
  • deleteKittyImage(...) / deleteAllKittyImages().
  • getImageDimensions / calculateImageRows / getCellDimensions / setCellDimensions.
  • An Image component or any maxWidthCells / maxHeightCells / imageId props.

If you need true in-terminal image rendering, it must be supplied by the host application; the framework only emits the text placeholder above.

Next Docs

Explore all productsDiscuss an implementation