model-status

name	model-status
description	Show current model/provider health, fallback chain status, and recent failures
version	1.0.0
author	system
tags	["models","health","monitoring"]

model-status

Invoke

/model-status          # Full dashboard
/model-status brief    # One-line per provider

Steps

1. Read model health state

cat /home/node/.openclaw/model-health.json

If the file does not exist, report: "Model health monitor has not run yet. The model-health-monitor hook may not be loaded. Check gateway logs."

2. Read current auth profile stats (for live data)

cat /home/node/.openclaw/agents/relay/agent/auth-profiles.json | jq '.usageStats'

3. Read recent notifications

tail -10 /home/node/.openclaw/model-health-notifications.jsonl

4. Format dashboard

Display using this format:

📊 Model Health Dashboard
Last checked: <lastChecked>

Provider Status:
  ✅ google — healthy
  ❌ anthropic — quarantined (billing) until <time>
  ⚠️ openrouter — rate-limited

Fallback Chain:
  1. google/gemini-3-flash-preview ✅
  2. google/gemini-3.1-pro-preview ✅
  3. openai-codex/gpt-5.3-codex ✅
  4. openrouter/auto ⚠️

Recent Events (last 5):
  [time] ❌ anthropic billing — credits exhausted
  [time] ✅ google recovered

Active: 3/4 models | Chain health: DEGRADED

Status icons:

• ✅ = healthy
• ⚠️ = rate-limited / cooldown
• ❌ = quarantined / disabled

If brief argument: one line per provider, e.g. google: ✅ | anthropic: ❌ billing | openrouter: ✅ | openai-codex: ✅

Notes

• Read-only. Does not modify any files.
• Uses data from model-health-monitor hook. If stale (>2min), warn user.