← BASELOAD English

Check Claude Server Status — Is It Down? A 1-Minute Check

June 18, 2026 · 51 min read 한국어 원문

To check Claude’s server status right now → the official status page status.claude.com is all you need. Below, we also cover the other methods (outage alerts, RSS, API, Downdetector). The traffic light above pulls from the official status API in real time.

When you’re deep in work with Claude and the screen suddenly freezes or you can’t connect, one thought comes first: “Is it my internet, or is the Claude server down?” Figuring this out within a minute saves you the wasted effort of needlessly rebooting your router or clearing your cache. This article walks through how to check Claude’s server status (whether there’s an outage) in real time, in order from the easiest method to the developer-grade ones.

Photo by Luke Chesser on Unsplash

The Most Reliable Place: The Official Status Page

The first place to look is Anthropic’s official status page.

• URL: status.claude.com (the old address status.anthropic.com also redirects to the same page)

The reason this matters is that Claude isn’t a single service — it’s broken out and displayed as several components. In other words, you can tell at a glance even in a situation where “the claude.ai web is down but the API is fine.” The six components currently published are:

• claude.ai — the web/app chat we all use
• Claude API (api.anthropic.com) — the developer API
• Claude Code — the coding tool for the terminal/IDE
• Claude Console (platform.claude.com) — console, billing, key management
• Claude Cowork — collaboration features
• Claude for Government — for government use

Next to each component, the status is shown by color. Read the meanings like this:

If “All Systems Operational” is showing at the top of the page, it means things are fine on Claude’s side — a signal that the problem is in your own environment.

If You Don’t Want to Check Every Time: Subscribe to Alerts

You can set it up so that you get notified first when an outage occurs, without even having to pull up the status page. Click the Subscribe button in the top-right of status.claude.com, and you can receive incident alerts through these channels:

• Email
• SMS (text)
• Slack
• Microsoft Teams

If you’re a solopreneur or developer who relies heavily on Claude for work, I recommend setting up a Slack or email subscription once. You’ll be automatically notified of when an outage starts and when it’s resolved, so you’ll stop wasting time wondering “why isn’t this working?”

Tracking History and RSS

When you’re thinking “this happened last week too, and now again?”, check the incident history.

• Past outage history: status.claude.com/history
• RSS/Atom feed: https://status.claude.com/history.rss — register it in an RSS reader and new incidents come in automatically.

There’s been a recent trend of more frequent brief delays and partial outages, so skimming the history page once will give you a sense of “whether today’s is a one-off problem or something that’s been recurring for days.”

Photo by Stephen Dawson on Unsplash

For Developers: Checking Status in Code

The status page is built on Atlassian Statuspage, so you can query the status directly via a JSON API. It’s easy to wire into a monitoring script or dashboard.

# One-line summary of the overall status
curl -s https://status.claude.com/api/v2/status.json

# Example response
# {"status":{"indicator":"none","description":"All Systems Operational"}}

The meanings of the indicator values are as follows:

• none → Normal
• minor → Minor issue
• major → Significant outage
• critical → Critical full outage

If you need per-component details or ongoing incidents, use /api/v2/summary.json and /api/v2/incidents/unresolved.json. If you want to automate further, you can register a webhook to receive incident create/update/resolve events directly on your server.

Put a Status Light in the Claude Code Statusline

If you build with Claude Code, you can wire the JSON API above into the statusline and see the server status as a 🟢/🔴 light the whole time you’re coding. No need to open the status page separately — just keep “Claude server 🟢 Operational” pinned to the bottom of your terminal.

The statusline is a custom status bar at the very bottom of the Claude Code terminal that shows things like the current model, directory, remaining context, and cost in a single line. Here we slot the status API into it.

The easiest way — the /statusline command. In Claude Code, describe what you want in plain language and Claude writes the script into ~/.claude/ and wires up the config for you (just approve the file edit):

/statusline show the current model plus a green/red light for Claude server status from the status.claude.com API

To set it up manually, add a statusLine block to ~/.claude/settings.json:

{
  "statusLine": {
    "type": "command",
    "command": "~/.claude/claude-status-line.sh",
    "padding": 2,
    "refreshInterval": 30
  }
}

Then save the script below as ~/.claude/claude-status-line.sh (make it executable with chmod +x). Since the statusline runs on every interaction, the key is to cache the status for 30 seconds so you don’t hit the network each time:

#!/bin/bash
input=$(cat)                                        # Claude Code passes JSON on stdin
MODEL=$(echo "$input" | jq -r '.model.display_name // "Claude"')
SID=$(echo "$input"   | jq -r '.session_id // "default"')
CACHE="/tmp/cc-status-$SID"; MAXAGE=30              # refresh only every 30s

stale(){ [ ! -f "$CACHE" ] || [ $(( $(date +%s) - $(stat -f %m "$CACHE" 2>/dev/null || stat -c %Y "$CACHE") )) -gt $MAXAGE ]; }

if stale; then
  curl -s --max-time 2 https://status.claude.com/api/v2/status.json \
    | jq -r '.status.indicator // "unknown"' > "$CACHE" 2>/dev/null || echo "unknown" > "$CACHE"
fi

case "$(cat "$CACHE" 2>/dev/null)" in
  none)           LED="🟢 Operational" ;;
  minor)          LED="🟡 Minor issue" ;;
  major|critical) LED="🔴 Outage" ;;
  *)              LED="⚪ Unknown" ;;
esac

echo "[$MODEL] Claude server $LED"

Now, the whole time you code, the bottom of your terminal shows:

[Opus] Claude server 🟢 Operational

The JSON the statusline script receives on stdin includes model.display_name (model), workspace.current_dir (current folder), context_window.used_percentage (context usage), cost.total_cost_usd (session cost), and more — so you can add the model, cost, or git branch next to the status light and build your own bar. (For the full field list, see the official Claude Code statusline docs.) When responses suddenly stall, just glance to the bottom of the screen and you’ll instantly see whether it’s your problem or a server outage.

Backup Options: Downdetector and X

In the “first few minutes” before the official page has reflected an outage, user-reported, lived experience is sometimes faster.

• Downdetector: Search “Claude” and you can confirm perceived outages via the spike-in-reports graph.
• X (Twitter) @AnthropicAI: During large-scale outages, the official account or user tweets are sometimes the fastest signal.

That said, these are nothing more than “perception” indicators. Make it a habit to confirm the final verdict with the official status page.

When It Still Won’t Work: A 30-Second Self-Diagnosis

If the status page says “Operational” but it’s only failing for you, the problem is most likely on your end. Just run through these quickly.

1. Connect from a different device/network (try mobile LTE) → if it works, it’s your Wi-Fi/corporate network
2. Use an incognito window or clear the cache → fixes login/session tangles
3. Turn off VPN and extensions → ad blockers or VPNs often block it
4. Pinpoint which of claude.ai / Claude Code / API is failing → cross-check against the status page components

These four steps will almost always sort out “Claude outage vs. my environment problem.”

Wrapping Up

To sum up, the No. 1 source is always the official status page (status.claude.com). Set up an alert subscription once, and going forward you’ll be notified automatically instead of fumbling around wondering “why isn’t this working?” If you’re a developer, wire it into your monitoring via the JSON API, and when the official page is slow to reflect an outage, cross-check with Downdetector. The more you rely on Claude as a work tool, the more this one checking routine cuts your downtime by the minute.