Installer and doctor report a failure for an unused Claude Code integration #870

Summary

While upgrading the Basecamp CLI using the official installation script, the installer reported an issue with the AI agent integrations and suggested running basecamp doctor.

The doctor command then reported a failed Claude Code plugin check, even though I only use Codex and haven't installed or configured Claude Code myself.

Steps to reproduce

1. Upgrade an existing Basecamp CLI installation using:

   curl -fsSL https://basecamp.com/install-cli | bash

2. The installer reports an issue with the AI agent integrations and suggests running basecamp doctor.

3. Run basecamp doctor.

I believe I upgraded from 0.7.2 to 0.13.0, but I'm not entirely sure about the previous version. Unfortunately, I no longer have the exact installer output.

Actual behavior

basecamp doctor reports:

✗ Claude Code Plugin Plugin not installed
    ↳ Run: basecamp setup claude
✓ Claude Code Skill Linked
✓ Claude Code Plugin Version Version not tracked
✓ Codex Plugin Installed and enabled
✓ Codex Plugin Version Up to date (0.13.0)

14 passed  1 failed

There is a ~/.claude directory on my system, possibly created by another application, but I haven't intentionally installed Claude Code or configured it for use with Basecamp.

Expected behavior

I wouldn't expect an unused, optional AI agent integration to cause an installation warning or a failed health check.

Since I only use Codex with Basecamp, the absence of a Claude Code plugin shouldn't be considered a problem.

Ideally, basecamp doctor would distinguish between integrations that are configured and those that are merely detected or available.