CLAUDE.md Service Table Still Stuck

The CLAUDE.md service table remains trapped in a loop of static text - ‘Running’ blinks on every entry, even as real logs show some services stuck or failed. This isn’t just annoying - it’s a false signal in a system built on transparency. #779 promised auto-correction, but the table hasn’t changed. The service status doesn’t match reality.

Here is the deal: the system treats ‘Running’ as a live status, not a stranded markup. SessionStart was supposed to fix this, but SessionStart hasn’t implemented the fix - launchctl still shows erratic exits, and #902 proves no plist exists to monitor true state.

Behind the scenes: users trust the table as a dashboard, but it’s a ghost of past fixes. The real question isn’t just why it’s stuck - it’s what we’re being asked to believe. Does ‘Running’ mean working? Or just a label clinging to outdated assumptions?

Three hidden truths:

• ‘Running’ in CLAUDE.md is a label, not a status - like checking a clock that never updates.
• The real health comes from launchctl logs, not static text.
• Retiring #779 without fixing the feedback loop creates quiet confusion.

The fix is simple: replace ‘Running’ with ‘launchd’ or ‘cron’ in the table, or better, let the DD reflect actual process status. Transparency isn’t just nice - it’s essential. When the table finally tells the truth, trust follows.