MessageData

The kind of message (user, assistant, tool, etc.).

Primary text content of the message.

For most message types this is the display text. For TOOL messages it is typically empty because the tool's identity comes from tool_name / tool_args instead.

Unique identifier used to match the dataclass to its DOM widget.

Unix epoch timestamp of when the message was created.

Name of the tool that was called.

Arguments passed to the tool call.

Current execution status of the tool call.

Output returned by the tool after execution.

Whether the tool output section is expanded in the UI.

User-supplied reason attached to a HITL reject decision (if any).

File path associated with the diff (DIFF messages only).

Name of the skill that was invoked.

Short description of the skill.

Origin of the skill (e.g., 'built-in', 'user', 'project').

User-provided arguments to the skill invocation.

Full SKILL.md content sent to the agent.

Whether the skill body is expanded in the UI.

Whether the message is still being streamed.

While True, the corresponding widget is actively receiving content chunks and should not be pruned or re-hydrated.

For APP messages, whether content is a markdown source string.

When True, rehydration renders the content via Rich markdown instead of the plain dim-italic AppMessage styling.

Cached widget height in terminal rows for scroll position estimation.

When _hydrate_messages_above inserts widgets above the viewport it needs to adjust the scroll offset so the user's view doesn't jump. Currently this uses a fixed estimate (5 rows per message). Caching the actual rendered height here after first mount would make that estimate accurate, especially for tall messages like diffs or long assistant responses.

Not yet populated — see _hydrate_messages_above in app.py.

Recreate a widget from this message data.

Create MessageData from an existing widget.