The dashboard
What each view does, and the keyboard shortcuts.
The dashboard is organised by workspace, then by project. Everything below is
reachable from the left rail; ⌘K (or Ctrl+K) searches the whole of it.
Issues
The feed is built for working through a list, not just reading one:
- Query tokens in the search field —
is:resolved,level:error,project:web— plus filters for project, group, level, assignee and date, and saved views you can name and reuse. - Bulk actions: tick rows (or the header checkbox) and resolve, archive or clear them together.
- Cursor paging, so the page continues from where the last one ended even while issues move.
- Row actions on hover: resolve/reopen, archive, mute for 24 hours.
- Live — the toggle above the list refetches every five seconds while it is on.
Keyboard shortcuts
| Key | Action |
|---|---|
j / k (or ↓ / ↑) | Move between issues |
↵ | Open the focused issue |
x | Select it, for the bulk actions |
e | Resolve, or reopen when resolved |
a | Archive, or restore when archived |
/ | Jump to the search field |
? | The shortcut card |
esc | Clear the focus |
Keys are ignored while a field has focus, so typing a query is never interrupted.
Explore
Subviews over the same query state — project, environment, time range and search. What you pick in the sidebar is in the URL, so a view is linkable. Traces, Logs and Metrics are live today; Profiles and Replays say what they need before they can show anything.
Traces
The trace list shows one row per transaction with its span and error counts. Open a trace for its waterfall:
- Each span can be opened for its own detail: operation, status, offset from the
trace's start, duration, self time (what it spent on its own work rather than
waiting on children), and its attributes — grouped by namespace, with the type
of every value beside it, so
"42"and42are not the same thing. - Subtrees collapse, and a collapsed row says how many spans it is hiding.
- Logs for this trace links to the log lines that belong to it.
Logs
Level tabs (all / trace / debug / info / warn / error / fatal) filter through the API rather than hiding rows client-side, rows expand to show their timestamp, trace, span, release, environment and attributes, and Load more grows the page.
The volume chart
Hovering a bucket names it — its window, and the counts per severity (or the
average, in latency mode). Clicking one opens that exact window for the whole
view; the chip beside the title returns to the preset range. Deliberately, the
chart never mixes units: selecting avg(duration) plots latency instead.
Project overview
The project's landing page: events and errors with a change against the previous window of the same length, crash-free sessions from the session rollup, unresolved issues, volume by severity, recent issues (with the same quick actions as the feed) and latency — the slowest transactions by p95, each with its p50, run count and error rate, linking into Explore filtered to that transaction.
A project's own settings — name and platform, DSN keys, source maps, team access — live under Settings in the project tabs.
Workspace
- Analytics — the last seven days across every project: issues, events, errors and the busiest issues.
- Dashboards — widgets you arrange, or generate from a prompt.
- Monitors — checks that report in.
- Settings — Profile (API tokens), Members, Teams, Alerts, Agent keys, Billing, Storage (rows and bytes held per project, so retention is a decision you can make with numbers) and the Audit log.
Small things that save clicks
⌘Kopens the palette: pages, projects and issues by name, with↑/↓and↵.- Live mode is per tab and deliberately not in the URL — a shared link opens a page once rather than starting to poll for whoever opens it.
- A trace and its logs link to each other: the waterfall has Logs for this trace, and the logs view shows a chip for the trace it is filtered to.