Status. And what it looks like from your side when we are down.

Right now

Private beta.

BlastRadius was deployed to production on 25 August 2026 and reviews pull requests from there. It is still a private beta — installation is by invitation — and the header will say so until anyone can install it without being asked.

There is no live indicator on this page and that is deliberate. A dot rendered when the page was built says "operational" during the outage you opened it to ask about, and a real one needs a script and a request to somebody else’s server — neither of which exists anywhere on this site. When there is something to report, it is reported here in words and by email to affected installations.

What an outage does to your build

This is engineered rather than promised, and it is the part that does not change with uptime.

• It never turns your build red. Nothing in the Action sets a failing exit code — not a missing token, not a plan we refused, not us being unreachable. The loudest it gets is an annotation and a line in the job summary.
• No comment and no check run is what an outage looks like. If the upload cannot reach us, nothing is posted. A branch protected on the BlastRadius check will then wait rather than fail — the merge button stays disabled with the check pending, which is the symptom to recognise.
• A review we could not perform reports, and it does not block. When the plan reaches us but we decline to review it — a plan terraform could not produce, an estate over the size cap, an afternoon of ours — the check run posts as neutral, which passes a required check. Only a critical finding fails one.
• There is no retry, on purpose. The comment is derived from the plan, so a missed one costs you the next push rather than any data. A retry would write a second run for every flaky network and would still not be the review you wanted.

No comment on your pull request? Look here first

1. The job summary on your own Actions run. Open the workflow run and scroll to the summary. It names the reason in a sentence — a missing token, no id-token: write permission, a plan that failed to build, an API it could not reach. That is faster than this page and it is specific to you.
2. The troubleshooting table. Every grey check this product can post, the symptom on the check run itself, and the fix.
3. Then write to us. Include the link to the Actions run. security@blastradiusapp.com.

Maintenance and changes

A deploy is a rolling restart, so an upload in flight during one is the only thing at risk, and the answer to that is the previous section: nothing goes red, and the next push is reviewed. There is no maintenance window to announce.

A change that reduces what you get — a retention window, a limit, a promise — is emailed to affected installations before it takes effect. What is retained and for how long is on the data handling page.