sup-abide: statusline supervises abide health

[msuarz]

Statusline code detects and fixes abide failures every tick.

Health Checks

1. Watcher not running — detect dead await, restart abide
2. WIP without a todo — orphaned message, create task or reset to new

Plan

~/.claude/plans/sup-abide.md

[msuarz]

Done

Watcher detection

• Abide writes watcher PID to status.json via pgrep after await starts
• Statusline checks PID liveness with Process.kill(0, pid)
• Own dude: pgrep fallback if PID missing/dead — self-discovers and writes PID
• Other dudes: rely on their status.json only
• No PID = dead → red ˣ on statusline

WIP stuck detection + self-healing

• Statusline reads inbox for "wip" messages
• Finds active session via newest .jsonl in ~/.claude/projects/<encoded-cwd>/
• Reads tasks from ~/.claude/tasks/<session-id>/ looking for "Abide" subject
• Wip + no matching task = stuck → resets message to "new" so abide picks it up again

Refactoring

• Single parse of status.json and inbox.json per dude (was 2 each)
• Extracted read_json, pid_alive?, has_abide_task? helpers
• has_abide_task? uses any? for early exit

Tests

• 54 tests, all green
• Covers: dead PID, alive PID, pgrep fallback, wip stuck, wip with task, no wip, missing status.json