Loading…
Changelog Observer collects what vendors publish about their services, turns it into one kind of event, and gives you and your agent the tools to decide what to do about it. Here is what that looks like, page by page.
Every release, breaking change, deprecation, security advisory, announcement, incident and maintenance window, in one shape.
Each event carries the vendor’s own title and summary, a link to the source, when the vendor published it and when we first saw it, the vendor’s severity, the priority we gave it, and the rule that gave it. Events come in two families: status updates (incidents and maintenance) and changes (everything the vendor shipped or withdrew). The Overview shows them side by side; the Events page filters by family, category, severity, dependency and a free-text search.
Duplicates are folded away. When a status page posts five updates to one incident, you see one timeline with five updates, and only the first one can page you. Every event has a “why” page that explains, in sentences, how it was classified, what rule routed it, and what was delivered where.
Tables are yours to shape: drag a column to move it, drag its edge to resize it, hide the ones you do not need. The arrangement is remembered.
An incident has two lifecycles that are never merged.
Upstream is the vendor’s word: open, monitoring or resolved, with their severity and affected components. Ours is what you decided: unreviewed, investigating, affected or cleared, with the customer impact you have evidence for and a note saying what that evidence is. A vendor resolving an incident does not clear it for you; it stays open for you until a person or your agent clears it.
Jira Cloud, Linear, GitHub Issues, or an incident on your own Atlassian Statuspage.
Add the ticket system once under Rules & destinations, with the credential it needs. From then on every incident has a “Create ticket” control. The ticket carries both lifecycles, the project and dependency, the timeline’s timestamps, the vendor’s updates oldest first, your impact note, and a link back. Its closing line says that customer impact is never inferred from the vendor’s severity.
Rules say what reaches which destination: at once, in the digest, or not at all.
Without rules, sensible defaults apply: urgent and high-priority events go out immediately, the rest wait for the digest, marketing is dropped. Write your own rules on category, severity, priority, dependency, source and environment. A rule can also suppress. The precedence is simple: suppress wins, then immediate, then digest.
Each project has a private inbound address for the vendors that only notify by email.
Mail is parsed, classified by rules, and attached to the right dependency. The address is a routing id, not a secret, so the product checks who really sent each message.
What needs you first, then what changed upstream.
Open actions, failed deliveries and failing sources, the only items the product will not resolve on its own.
Drag sections between the top, the main column and the side rail, hide what you do not use, pick a density and a default range. Saved to your account.
Every project has an overview of its own with a shareable address; the workspace overview adds a scope selector and project cards.
Numbers you can trust the silence by.
Source coverage (how many polled sources succeeded in the last day, and which are stale), ingestion lag (from the vendor’s publish time to ours), alert delivery lag and delivery success, signal quality from your feedback, event volume by category and week, and time to first event per project. Over seven or thirty days, for the workspace or one project. Each number states its definition on hover, and the same numbers come back from the API, so your agent reports the same thing you see.
Owners invite by email; invitations are one-use links valid for seven days. Owners administer, members do the work.
Every change, whether a person made it in the app or an agent made it through a key, is recorded with who, what and when.
The free plan’s limits are published and your meters are visible, so nothing is a surprise.
Sign in, create a key, and ask your agent to set up the first project.