Why does Anthropic's status page say all systems operational when Claude Code is down?

Anthropic publishes several separate components — claude.ai, the Claude API, Claude Console, Claude Code and others — and the headline indicator summarises all of them at once. A failure confined to Claude Code can leave the overall indicator green. Status pages are also updated by humans once an incident is confirmed, which lags the first user reports by minutes to half an hour.

Can Claude Code be down while claude.ai works fine?

Yes, and it happens regularly. They are listed as separate components for exactly that reason. The chat interface working is not evidence that the CLI or API is healthy, and vice versa.

Anthropic status vs Claude status: why they disagree

The official page is not lying to you. It is answering a different question.

A familiar moment: Claude Code is plainly broken, you open Anthropic's status page hoping for confirmation, and it reads All Systems Operational. That mismatch has three separate causes, and knowing which one you are looking at tells you how long to wait.

Cause 1: the headline indicator averages everything

Anthropic publishes several distinct components, and they fail independently:

claude.ai — the chat interface.
Claude API (api.anthropic.com) — what most integrations call.
Claude Code — the coding agent, listed in its own right.
Claude Console (platform.claude.com) — billing, keys and settings.
Claude Cowork and Claude for Government.

The single indicator at the top summarises all of them. A failure confined to Claude Code barely moves that average, so the headline stays green while your CLI stays broken. This is the single most common reason people conclude the status page is wrong — it is not wrong, it is just aggregated.

The practical move: ignore the headline and read the per-component list, which is exactly what the service breakdown on our front page shows.

Cause 2: status pages lag reality

Every status page in the industry works the same way: an incident is posted once a human has confirmed it. Between the first failing request and the first status update there is a gap — routinely ten to thirty minutes, and longer for partial degradations that are hard to distinguish from noise.

So “the status page says it is fine” is weak evidence in the first half hour of an incident and strong evidence after it. Five minutes in, trust your own failing requests over the green tick.

Cause 3: it genuinely is not them

The uncomfortable third option. If the official feed is green, we are green, and your requests still fail, the cause is very likely local — an expired session, a usage limit, a proxy, or a hanging MCP server. We have a page for each: CLI troubleshooting and usage limits.

Which should you trust?

For “can I work right now?” — a plain verdict like this one, plus the per-component breakdown.
For “what actually broke, and is there an ETA?” — Anthropic's official page, because incident write-ups and resolution notes only exist there.
For “is this normal?” — our outage history, which counts how often this has happened before.

They are complements, not competitors. We read Anthropic's feed — we simply present it as a single answer to a single question instead of a dashboard.

A note on what we are not

This is an independent, unaffiliated project. We have no privileged information, we do not probe Anthropic's infrastructure, and we cannot tell you when an incident will be resolved. Everything here derives from the same public feed you can read yourself — the value we add is the plain-English verdict and the email alert, not inside knowledge.

Not affiliated with, endorsed by, or sponsored by Anthropic. Claude and Anthropic are trademarks of Anthropic. Status is read from the official status.claude.com feed.