Node-based terminal manager for AI coding agents — tmux-backed terminals and parallel agent sessions as draggable nodes on an infinite pan/zoom canvas. macOS, Linux, and a browser Server Edition.
TypeScript adhdagent-orchestrationai-agentscanvasclaude-codecodexcoding-agentsdeveloper-toolselectrongeminilinuxmacosmultiplexeropencodeparallel-agentsterminalterminal-emulatorterminal-multiplexertmuxworkspace-manager
fix(chat): render single line breaks and quotes as Claude Code does
Paul Robertson63ec77c11680
diff --git a/src/renderer/nodes/ChatPanel.thread.test.tsx b/src/renderer/nodes/ChatPanel.thread.test.tsxindex 643d344f8..1cb5fe518 100644--- a/src/renderer/nodes/ChatPanel.thread.test.tsx+++ b/src/renderer/nodes/ChatPanel.thread.test.tsx@@ -84,6 +84,19 @@ describe('⌘M thread look', () => {expect(assistant.querySelector('.term-chat__text strong')?.textContent).toBe('now')})+ it('keeps single line breaks, as Claude Code renders them — in prompts and in quoted replies', async () => {+ await mount([+ { role: 'user', parts: [{ kind: 'text', text: 'This is a test\nof a multiline message' }], key: 0 },+ { role: 'assistant', parts: [{ kind: 'text', text: '> This is a test\n> of a multiline message' }], key: 10 }+ ])++ const user = host.querySelector('.term-chat__msg--user .term-chat__text')!+ const quote = host.querySelector('.term-chat__msg--assistant .term-chat__text blockquote')!++ expect(user.querySelector('br')).not.toBeNull()+ expect(quote.querySelector('br')).not.toBeNull()+ })+it('one action row per assistant TURN (not per tool line), the latest always shown', async () => {await mount(thread)const rows = [...host.querySelectorAll('.term-chat__actions')]diff --git a/src/renderer/nodes/ChatPanel.tsx b/src/renderer/nodes/ChatPanel.tsxindex cbe4247f0..37e95ad3b 100644--- a/src/renderer/nodes/ChatPanel.tsx+++ b/src/renderer/nodes/ChatPanel.tsx@@ -36,8 +36,10 @@ import { assistantTurnEnds } from '../lib/chatThread'// Memoized bubble: marked+DOMPurify re-ran for EVERY message on each ChatPanel render (each// turn-finish reload, each keystroke re-render). Text is stable per message, so cache per text.+// `breaks`: a single newline is a line break, as Claude Code's own TUI renders it — without it a+// multi-line prompt, or a reply quoting one line by line (`> a` / `> b`), collapsed onto one line.export const MarkdownText = memo(function MarkdownText({ text }: { text: string }) {- const html = useMemo(() => renderMarkdown(text), [text])+ const html = useMemo(() => renderMarkdown(text, { breaks: true }), [text])return <div className="term-chat__text" dangerouslySetInnerHTML={{ __html: html }} />})diff --git a/src/renderer/styles.css b/src/renderer/styles.cssindex cb660db8b..eab592504 100644--- a/src/renderer/styles.css+++ b/src/renderer/styles.css@@ -3407,6 +3407,14 @@ body,text-decoration: underline;text-underline-offset: 2px;}+/* Same treatment as the markdown view's quotes (.term-md__content blockquote); without it a quote+ in the thread was only the browser's default indent. */+.term-chat__text blockquote {+ margin: 6px 0;+ padding-left: 10px;+ border-left: 3px solid var(--border);+ color: var(--muted);+}/* A tool call in the unbubbled assistant flow: a quiet chip on the panel itself (a light tint and ahairline), not a dark box nested in a bubble that is no longer there. */.term-chat__tool {