Is it down?
Slack won’t load, the Zoom call keeps dropping, GitHub times out. Kilobat checks your own connection and the service’s public status page, then says which side it’s on, like “Slack is down for everyone, not you.”
brew install seifip/kilobat/kilobat$ kilobat down
✗ Slack Slack is down for everyone — Messages are failing to send
https://slack-status.com
✓ Zoom Zoom’s status page reports no problems.
✓ GitHub GitHub’s status page reports no problems.Kilobat
Slack is having an outage
Messages are failing to send. Slack kept failing to connect while your own connection looked fine.
A status page alone can’t say whether your Wi-Fi is the problem, and a speed test can’t say whether Slack is. Kilobat looks at both, and only repeats what the status page says.
Your Wi-Fi, router, provider and DNS, then that app’s status page.
“Slack is down for everyone, not you”, with the incident’s title and a link to the status page. Free
Connections that keep failing to open, while your own line is fine.
One notification per incident, such as “Slack is having an outage”, with a button that opens the status page. Turn it off in Settings. Free
Freezes beyond your router: where the calling service’s own servers are.
The call’s card says whether Zoom’s or Teams’ status page reported an outage at the time. Free
The path from your Mac to that one site.
“Slack’s status page reports no problems”, then the layer that broke: DNS, a VPN or proxy, IPv6, the server. Free
Why can’t it connect?
Name a site or an app and Kilobat checks your Wi-Fi, DNS, VPN and proxy, its own rules, the connection and the server’s answer, then asks that service’s status page. When the page reports an outage and nothing on your side failed, the verdict leads with it. When your Mac is the problem, it says so first.
# Is it them? The status pages of the apps running now
$ kilobat down
# One service, by name or by site
$ kilobat down slack
$ kilobat down github.com
# Or me? Every layer from this Mac to the server, and its status page
$ kilobat why app.slack.comThe app’s proxy isn’t answering.
Why can’t it connect? Kilobat tests the request and path the app itself uses, its own proxy, IPv6 and tunnels included, and names the layer that broke.
Kilobat knows which apps and sites belong to which service, by bundle ID and domain, so a failing Slack app or a slow github.com points at the right page.
Some services publish no public status, so Kilobat can’t check them: Webex, Teams for work or school (shown to admins only), Spotify, Steam, WhatsApp, Telegram and Signal.
When Claude Code stalls on errors or Codex keeps retrying, Kilobat checks your line and the provider’s status page, for the part your agent actually uses: Claude Code, not claude.ai; Codex, not ChatGPT; Copilot, not the rest of GitHub. If the agent keeps failing to connect while your line is fine and the provider reports trouble, you get one notification: “Claude Code: Anthropic reports an outage — not your connection.” Any app talking to api.anthropic.com, api.openai.com or another provider’s API is noticed the same way, and agents can ask for themselves through the MCP tool service_status.
Not covered, because there’s no public feed Kilobat can read: Grok and the xAI API, DeepSeek, OpenRouter, Replicate, JetBrains AI, Kiro and Amazon Q, and the Gemini API in Google AI Studio. Ollama and LM Studio run on your Mac, so there’s nothing to check. Aider and other agents that use your own choice of provider are matched by the API they connect to.
Health
Open Health when things feel slow. If your own line checks out, Kilobat asks the status pages of the apps you’re using and puts any outage first, so you don’t restart the router for nothing. When they’re all green, it says that too. While a service you use has a problem, a Services row appears in the menu bar menu; when all is fine, it isn’t there.
Getting warmer. Keep going.
DNS lookups are slow. Compare Cloudflare or Google DNS, then restore Automatic if it does not help.
kilobat down, kilobat why and the MCP tool service_statusPro is $19 once for two Macs, with a 7-day free trial and 14-day refunds. Free for students.
It reads that service’s own public status page, the one its team updates, such as slack-status.com or githubstatus.com. It only repeats what the page says, and says “down for everyone, not you” only when the page reports an outage and your own line checked out.
Downdetector counts reports from other people. Kilobat checks your own Wi-Fi, router and DNS first, then the service’s official status page, so it can tell you which side of the problem you’re on.
No. It sends a plain request to the status page, with no cookies and nothing about you, your network or your apps. It asks only when there’s a reason, and remembers each answer for three minutes. Read the privacy notice.
No. It asks when you open Health or check a connection, when a call froze beyond your router, and when an app with a status page keeps failing to connect. Otherwise nothing is sent.
Yes, all of it, including the notifications and the call cards. Kilobat Pro, $19 once, adds findings with a fix for your own network, the cause of each call freeze and history.
Yes. kilobat down --json and the MCP tool service_status return the same answer, and diagnose_connection includes it for a host it checks. See the docs.
brew install seifip/kilobat/kilobat