Incident detail workflow

Incident detail is where active response turns scattered signals into a single operational story. Use it to collect the facts, decide on the next action, and leave a record that other teammates can understand later.

Start with the summary

Begin with severity, title, owner, status, and confidence. Confirm the incident still describes the current impact before diving into individual evidence rows.

Review evidence

Evidence rows collect signals such as error groups, logs, uptime failures, performance changes, infrastructure saturation, deploys, and integration activity. Read them in time order to understand what changed first and what followed.

• Use uptime evidence to confirm customer-visible availability impact.
• Use deploy diffs and source context to identify likely changed code.
• Use logs and infrastructure evidence to distinguish application defects from platform pressure.

If the incident needs deeper context, open the linked logs, performance traces, infrastructure dashboards, uptime monitors, or session replays from the same time window.

Use AI investigation

AI investigation output can summarize likely causes, confidence, and proposed next steps. Treat it as a fast analysis aid, not as the final incident record. Verify important claims against the linked evidence before acting.

Check replay and source

Replay sessions and source context help connect technical symptoms to user behavior and code paths. Use replay to understand what happened before a browser or request failure, and source context to confirm whether a proposed fix targets code your team owns.

Link repair work

Link the incident to issues, pull requests, or repair-agent output when implementation work leaves Errorgap. The incident should still retain the operational summary, impact, and key evidence so responders do not need to reconstruct the event from another tool.

Compare similar incidents

Similar incidents can reveal recurring causes, incomplete fixes, or seasonal patterns. Compare affected projects, releases, evidence types, and final resolutions before declaring the current incident unique.