Skip to main content

Incident detail

Open an incident and this is what you get: the full write-up, the badges, and the buttons that turn a problem into tracked work.

Where to find it: click any card on the Incidents list (/operation/incidents). The detail view lives at /operation/incidents/:projectId/:id.

Two older URLs still work: /monitoring/:projectId/incidents/:id redirects straight here, and /monitoring/incidents/:id (no project in the URL) opens the same page, falling back to whatever project was passed along or is currently active.

A Back link (arrow + "Back") sits above the page and returns to whatever page you came from.

Loading and error states​

  • While the incident is being fetched: a spinner with "Loading incident…".
  • If the incident cannot be found (bad ID, wrong project, or it was removed): a shield icon, "Incident not found", and a Back to Monitoring link to /operation/incidents.

The header​

  • Eyebrow: the incident's source ID.
  • Title: the incident's display title. Normally this is the incident's own title. If the raw title is actually a serialized JSON payload (happens when the ingest pipeline could not extract a readable title), ACE falls back to a PR title if one is in the payload, then to the incident's summary, and if none of those exist, the title shows literally as "Evento sem título" (this fallback string is not translated).
  • Description: the incident's summary, or its raw description if there is no summary, or "No description provided." if neither exists.

Action buttons appear on the right of the header, depending on what is linked and what you are allowed to do:

  • Open issue: shown when an issue is already linked to this incident and you can read issues. Takes you straight to that issue's card.
  • Create issue: shown instead, when there is no linked issue yet and you are allowed to manage incidents and create issues (and the project is not read-only). Opens the Create issue modal described below.
  • Neither button shows if you do not have the right permissions, or if an issue is linked but you cannot read issues.
  • Discuss in chat: shown if you have chat access. Opens (or creates) a chat conversation for this incident and navigates to it. If it fails to open, a red error line appears under the header: "Could not open this incident conversation."

What do the badges mean?​

A card just below the header shows one or more badges, always in this order:

  • Severity: the incident's severity (for example Critical, High, Medium, Low). Critical renders in the loudest, highest-contrast style; High in red; Medium in yellow or amber; anything else in a neutral style.
  • Status: the incident's status. Resolved and closed statuses render green; fixing, resolving, investigating, and acknowledged render yellow or amber; open renders blue; anything else neutral.
  • Type: the incident's type (for example "Security Issue", "Cost Anomaly"), if it has one.
  • Source: where it came from, if known.
  • ×N occurrences: shown only when this incident has fired more than once. Rising occurrence counts mean the same underlying problem keeps happening.
  • flapping ×N: shown only if the incident has a non-zero flap count, meaning the problem has started and stopped repeatedly.
  • merged: shown only if this incident was folded into another one (you would normally only land here through a direct link, since merged incidents do not appear in the list).

Below the badges, the incident's raw description is shown as plain text, but only when it exists and adds something beyond the summary already shown in the header (it is hidden if it is identical to the summary).

What is the Agent analysis panel?​

A second card, headed "Agent analysis" (with a sparkle icon), shows ACE's write-up of the incident:

  • If you do not have chat access, it shows "Chat access is required to view the incident analysis thread."
  • If you have access but ACE has not finished analyzing yet, it shows a spinner with "ACE is analyzing this incident…". Analysis runs asynchronously right after the incident is detected, so this can take anywhere from under a minute to a few minutes; the page polls in the background and the write-up appears as soon as it is ready.
  • Once ready, the analysis renders as formatted text (Markdown). This is the same conversation thread that Discuss in chat opens, so anything you ask there continues from this analysis.

What's in the sidebar​

A narrow column on the right shows, section by section:

  • PROJECT: the project's name (or its raw ID if the name is not available).
  • AGENT: the agent involved, if the incident has one assigned. Hidden if none.
  • LINKED ISSUE: only shown if an issue is linked. If you can read issues, shows the issue's title (or key) and its status. If you cannot, shows a generic note: "An issue is already linked to this incident."
  • DETECTED: when the incident was first detected.
  • UPDATED: when it was last updated.
  • INGESTED: when it was ingested into ACE.
  • RESOLVED: only shown once the incident has a resolved timestamp.

All dates are shown as "Mon D, YYYY, H:MM AM/PM"; any missing date shows as "—".

Create issue modal​

Opened by clicking Create issue in the header (only available when there is no linked issue yet and you have the right permissions). It turns the incident into a card on the Development board and starts an agent on it right away: this is the bridge from an operational problem to an actual fix.

The dialog is titled "Create issue from incident", with a Close (✕) button in the top corner, disabled while the issue is being created.

Fields:

  • Issue title: a text field, pre-filled with [<source ID>] <incident title>. Editable. Creation is blocked if this is left empty.
  • Agent brief: a multi-line text box, pre-filled with a Markdown brief built from the incident: an "Incident Details" section listing source ID, severity, status, type, source, and detected time (whichever of these the incident has), followed by a "Description" section with the incident's raw description, if it has one. It is fully editable, so rewrite it if you want the agent to focus on something specific.
  • Repository (optional): a dropdown showing the project's repositories by full name, with the default branch shown underneath the selected one. If the project has no repositories connected, it shows "No repositories" and the dropdown is inactive. Leaving this unset is fine, as the agent falls back to the project's bound repository if one exists.
  • A status line below the fields: "Starts In Progress — agent dispatches immediately," telling you the resulting card will not sit in a backlog.

Footer button: Create & dispatch, disabled until the issue title is non-empty (and while a project, and the incident, are known). While saving it shows a spinner and reads "Creating…". On success the modal closes and the header now shows Open issue instead. If creation fails, a red error line appears above the button with the server's error message, or "Could not create the issue." as a fallback.

The new card carries the incident's own analysis thread over as its conversation: the same thread you were reading under Agent analysis continues on the card.

On this page