Add Claude and Codex desktop application diagnostics

[nyldn]

Desktop-only installations currently appear to lack Claude or Codex because diagnostics only look for terminal commands. Add separate macOS desktop bundle detection, keep missing standalone CLIs informational when a desktop app is verified, and explain which tasks still require a CLI or an in-app check.

The new check validates known bundle identities under system and user Applications folders, recognizes the Codex identity under both Codex and ChatGPT names, deduplicates aliases, and reports distinct app copies. It extracts only allowlisted plist fields and never launches app-contained runtimes. Manual Claude Desktop MCP configuration has a separate redacted inventory, with explicit limits for extensions, remote connectors, and runtime health.

Research uses current fetched Anthropic and OpenAI documentation. The guide distinguishes Claude Chat, Cowork, local Code, and cloud execution; documents current desktop MCP sharing; and records the OpenAI documentation redirect to ChatGPT desktop. The earlier browser-routing instruction commit is included so desktop users get the same agent contract.

Validation:

• macOS system Ruby: 101 tests, 897 assertions, no failures.
• Ruby and shell syntax checks passed.
• Release archive built from the committed revision; isolated install/uninstall, tamper rejection, and legacy upgrade checks passed.
• Gitleaks history and working-tree scans passed.
• Live read-only doctor and agent-context verified desktop and CLI findings independently.
• Independent code review found a legacy ChatGPT false warning; the fix and regression assertions passed focused review with no remaining findings.

This changes diagnostic code and documentation only. It does not install provider CLIs, alter provider configuration, authenticate accounts, start apps, or change schedules. Custom app names and installation locations remain outside the bounded scan. App metadata is installation evidence, not authentication, signing, entitlement, or connection-health proof.

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Team

Run ID: be7d9cea-1fbe-4fc2-bc36-4ea8b20a2b32

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}