Watchdog monitors the operational and security environment around a registered release. Credible evidence reaches the Medic. The Medic decides whether the Council should investigate.
An audit is a photograph. It tells you the state of the work on the day it was taken. The day after you ship, a dependency you rely on gets an advisory, a package in your tree is compromised, or a failure mode nobody modelled starts appearing in production.
Nothing in the original audit is wrong. It has simply stopped being current, and there is no mechanism that notices. Watchdog is that mechanism, and the Council is what it reports to.
Eight stations, and the eighth returns to the first. That is the point: there is no end of the process, because there is no point at which the world stops changing around your release.
Watchdog detects. The Medic diagnoses. The Council decides. You release.
Not a generic web search. A watch built around one registered build, its runtime, its frameworks and its actual dependency inventory.
Participating applications send constrained, sanitised incident evidence. Not arbitrary private logs.
Repeated failures become one stable incident, not hundreds of alerts.
Trusted security intelligence is correlated against the components actually present in your release.
Specific to this application, this build, and its actual dependencies.
A vulnerable component does not prove exploitability. It creates an evidence-backed obligation to investigate, which is a different and more useful thing than an alert.
Detection on its own produces noise. The Medic is the Council role that receives Watchdog's evidence and decides whether it warrants the Council's attention at all.
It weighs severity, urgency, exposure, recurrence and which releases are affected. Most things stop here, which is the job working correctly. Unresolved questions stay visible rather than being closed for tidiness.
When action may be warranted, the Medic puts a structured case to a council composed for that specific problem.
Watchdog detects. The Medic diagnoses. The Council decides. You release.
Four parties, four jobs, and no step where a machine both proposes a change and approves it. The last word is a person's, and that is not a limitation we are apologising for. It is the design.
Software that can recognise danger, call for expert review and come back stronger. None of which requires it to act on its own.
Hosted monitoring runs around the clock. Cloud compute is metered and paid by you as a small operating cost, at your provider's price.
No machine of yours needs to stay powered on.
Run Watchdog on your own infrastructure with your own provider keys. Nothing leaves your estate that you did not send.
The hosting machine must stay powered and connected for monitoring to continue.
Named rather than hidden. You can see the behaviour and the evidence without being able to rebuild the mechanism.
It watches what a council already reviewed, and brings anything that changes back to that same council. It is not sold on its own, because on its own it would have nowhere to report.