Skip to content
Agents tracked: 502 Releases (7d): 762 Agent status: no reported issuesSince last limit reset: Claude 11d · Codex 9d For AI agents: Markdown Updated Nov 22, 2026

openstatus

by openstatusHQ

run status pages and uptime checks from chat.

OtherOfficialRemote

GitHubWebsite

1. Made by

“dev.openstatus/* is the reverse-DNS namespace for openstatus.dev.”

The repo openstatusHQ/openstatus is openstatus's own, the server runs on api.openstatus.dev and the docs on openstatus.dev describe it; the registry name dev.openstatus is proven by a DNS record on openstatus.dev.

github.com/openstatusHQ/openstatus/blob/bba13ddc9e304695547f039cff849ed7e9826c58/apps/server/MCP_REGISTRY.md

Checked against its repository at one commit, on Nov 10, 2026. Tap a verdict to see its source.How we checkedReport an error

What openstatus does

openstatus checks that your website or service is up and shows a public status page to your users. Its MCP server lets your AI assistant read those checks, post incident updates, schedule maintenance and set up new pages from a chat. You sign in once in the browser and pick what it may change.

What people use it for

Tell users about an outage fast

When something breaks, ask the assistant to draft a public status update. It shows you the draft and asks whether to email subscribers before posting.

Find out what is failing

Ask "what is failing right now?" and the assistant reads each region's health and the latest check results, with status codes and timings.

Announce planned maintenance

Plan a database upgrade for Sunday night and let the assistant put the window on your status page, with or without telling subscribers.

Set up checks and a page for a new app

Launching a new site? The assistant adds an uptime check for its address, creates a status page and puts the check on it.

Run an incident and write the postmortem

Open an internal incident, add notes as you work, then let the assistant draft the postmortem from the timeline for you to approve.

A weekly uptime summary Idea

Every Monday, ask the assistant for each check's success rate and response times over the last 7 days and paste it into your team update.

openstatus popularity

9,274
GitHub stars, openstatusHQ/openstatus
0
public agent projects connect it
openstatus: GitHub starsopenstatusHQ/openstatus 9,274

Tools it gives your agent (50)

What it sends and its license

Sends

what you ask the assistant to read or post (check results, notices, maintenance windows, postmortems)

To openstatus (api.openstatus.dev); on every tool call. Opt out: self-host openstatus with your own OAUTH_ISSUER.

Sends

public updates and, when notify is true, messages to your status page subscribers

To your public status page and its subscribers; create_status_report, add_status_report_update, resolve_status_report, create_maintenance, add_maintenance_update. Opt out: pass notify false, or connect with Read only access.

Sends

every change, with the connection and the person behind it

To the workspace audit log; on every change made through the server. Opt out: none.

openstatus is open source under the AGPL-3.0, and the hosted service starts with a free Hobby plan. If you change its code and run it for others, you must share your changes under the same license.

In a paid app
Yes

Using the hosted service is allowed on any plan, starting with the free Hobby plan.

“Available as managed SaaS or self-hosted (AGPL-3.0).” source

Pass on its files
With conditions

The code is AGPL-3.0; changed versions you share or run as a network service must offer their source under the same license.

“GNU AFFERO GENERAL PUBLIC LICENSE” source

Is it alive

Latest version

1.0.0 in the registry, updated Sep 10, 2026

Last release

Registry entry 1.0.0 published 2026-09-10; code read at commit bba13ddc9e304695547f039cff849ed7e9826c58 (2026-11-10)

Remote endpoint
yes: an unauthenticated HEAD to https://api.openstatus.dev/mcp answered 401 (the server is there and asks for sign-in)

Alternatives

openstatus This page

9,274 starsRemote

oneuptime

Pick it if you also need logs, traces and metrics in the same product

Local

honeybadger

Pick it if error tracking inside your app matters more than public status pages

Local

All other servers

Get started

Tell your AI assistant

“Use the openstatus MCP server. List my monitors and tell me which ones are failing and in which regions; do not change anything.”

Connect it

claude mcp add --transport http --scope user openstatus https://api.openstatus.dev/mcp

Mistakes to avoid

Calling create_status_report without notify, or deciding notify yourself
Show the draft, ask the person whether to notify subscribers, then pass notify true or false

The field is required with no default, and a notice sent without notify can never be sent to subscribers later.

Typing a page, monitor or component id from memory
Look it up with list_status_pages, list_page_components or list_monitors first

Ids from outside the workspace or the page come back as NOT_FOUND.

Sending update_monitor with only the new alert channel in notificationIds
Read the monitor with get_monitor and send the existing ids plus the new one

Array fields replace the stored set, so a partial list unlinks the other channels.

Assuming resolve_incident also closes the public status report
Call resolve_status_report as well when the public notice is still open

Internal incidents and public reports are resolved separately.

Reading notified true as proof that every subscriber got the message
Treat it as "the sending step ran"; check delivery in the dashboard if it matters

The flag only says the dispatch call did not fail.

Questions

Do I have to paste an API key?

No. Interactive AI apps sign in through a browser consent screen; an API key is only the fallback for CI and servers.

How many tools does it have?

50, or 48 on plans without the audit log; 4 of them read public openstatus.dev pages.

How do I cut off the assistant's access?

Revoke the connection under Settings, Integrations, Connected apps; its next call fails.

Does declaring an incident publish anything?

No. Incident tools are internal; only status reports and maintenance tools publish to the status page.