fix(cli): preserve full URL targets in truncated box titles

[yangshun]

Description

Deployment URLs in narrow summary box titles lose their terminal hyperlink when the renderer truncates the title.

✓ built in 120ms
◇ Preparing deployment...
◇ Applying D1 migrations...
╭─ https://void-todo.yangshun.workers…─╮
│                                      │
│  Worker: void-todo                   │
│  D1 migrations up to date            │
│                                      │
╰──────────────────────────────────────╯

✓ Successfully deployed!

Clicking on it led me to https://void-todo.yangshun.workers… which is an invalid domain.

Fix

Keep the shortened title and attach the original full URL as its hyperlink destination.

I tried out another ways like shifting the URL into the box, but settled on this approach as it requires the minimal changes.

Validation

• Focused deployment summary and terminal hyperlink tests
• vp run test (affected tests)
• vp run --no-cache test:all (formatting, lint, builds, typechecking, and the full test suite)
• Packed React consumer smoke tests with npm and pnpm

Tested the following in my terminal, URL is truncated but clicking on it opens the full URL in the browser.

 FORCE_HYPERLINK=1 node --input-type=module <<'JS'
import { note } from './packages/void/src/cli/output.ts';

note(
  'Worker: todo-app\nD1 migrations up to date',
  'https://example.com/?app=todo-app&deployment=0123456789abcdef',
  { output: { columns: 40, write: chunk => process.stdout.write(chunk) } },
);
JS
╭─ https://example.com/?app=todo-app&…─╮
│                                      │
│  Worker: todo-app                    │
│  D1 migrations up to date            │
│                                      │
╰──────────────────────────────────────╯

AI disclosure

Used GPT-6.1 Sol Max. It's initial direction was superfluous and modified a lot of code. I decided on this approach which is IMO the smallest change needed to fix the issue.

Maintainer follow-up

Rebased onto main, with one commit on top: the cut title is found by matching the URL itself (its longest start followed by …) instead of the / (\S*…)/ regex, so no other word ending in … can be taken as the link label.

🤖 Generated with Claude Code