Vibeos Ops Health

Triage gateway, cron, and kanban health.

Skill metadata

Source	Bundled (installed by default)
Path	skills/devops/vibeos-ops-health
Version	1.0.0
Author	VibeOS
License	MIT
Platforms	linux, macos, windows
Tags	ops, gateway, cron, kanban, health
Related skills	git-local-workflow, issue-to-pr

Reference: full SKILL.md

info

The following is the complete skill definition that VibeOS loads when this skill is triggered. This is what the agent sees as instructions when the skill is active.

VibeOS Ops Health

Operator playbook for the personal-AI-OS moat: gateway liveness, messaging reconnect, cron ticks, and kanban dispatcher health. Prefer these CLI checks over guessing from chat silence.

When to Use

• Gateway "feels stuck" or chats stop answering
• Cron jobs missed a window
• Kanban ready queue not spawning workers
• After Wi-Fi / DNS blips

Prerequisites

• Local install with vibeos / ./scripts/vibeos on PATH
• Read access to ~/.vibeos/logs/ (or profile home)

How to Run

1. Gateway + reconnect queue:

vibeos gateway status

Look for ↻ platform: reconnect in Ns and ⚠ kanban dispatcher stuck.

2. Cron triage:

vibeos cron status
vibeos cron list

cron status now lists last-failed, delivery-error, and overdue jobs.

3. Kanban:

vibeos kanban list --status ready
vibeos kanban stats
vibeos kanban diagnostics

4. Logs:

vibeos logs --follow --level WARNING
# or: ~/.vibeos/logs/gateway.log

Quick Reference

Symptom	Check	Likely fix
No chat replies	gateway status	vibeos gateway restart
Platform reconnecting	status ↻ lines	Wait for backoff; fix token if fatal
Cron not firing	cron status ticker lines	Restart gateway; inspect job errors
Ready tasks idle	kanban list --status ready	Profile venv/PATH/credentials
Gave-up notify	message shows N/M failures	Raise kanban.failure_limit or fix spawn

Pitfalls

• Background delegate_task is process-local — use cron / terminal(notify_on_complete) for work that must survive restart.
• Prefer no_agent cron scripts for cheap maintenance (FTS hygiene) to avoid burning model turns.
• Do not add new core tools for ops — CLI + skills are enough.

Verification

• gateway status shows supervised PID and no unexpected fatals
• cron status shows fresh heartbeat and no overdue cluster
• Stuck kanban clears after spawn resumes (status no longer warns)