Everything under /app is the team's view of the hosted service. State is pushed over one server-sent events stream per session — a check landing from CI, a connection health result, a teammate's policy edit appear as they happen; nothing polls. What a member can change depends on their role (see Team below); viewers see every page and change nothing.
Checks — the run log
Every check that reached the service, from the Action, --remote or the API, with its files, findings, the rules that ran, the suppressions that were honored or refused, and the connection its context came from. Filter by repository once the log holds checks from more than one. A check page shows each finding with its rule, the SQL it fired on and the fix, and warns when the live-schema context it used came from a connection that is now unreachable, removed, or has not been reached in a day. Every finding has a dispute button: use it when the rule misread the SQL, not to accept a correct finding — accepted findings are muted with an inline comment, a repository ignore or the team policy instead.
Insights
The run log as trends: which rules fire most, how the clean-run rate moves, which days went sideways, which repositories carry the findings, which rules get disputed. Each rule links to its own page with its history on your team. The window is a plan lever — 30 days on Startup, a year on Team, unlimited on Scale — and it is a window over the run log, never a deletion.
Connections
Stored connections and protected databases, as described in the hosted-service page. Each connection shows its last health result — reachable, table count, server version — from a probe every 15 minutes and on every use. A connection whose role could write (superuser, table owner, or a write grant) carries a warning chip pointing at the catalog-only role from the setup guide: Bolvrk only reads catalogs, so a writable role is exposure with no benefit. Startup and above.
Policy
One place to decide what fails a build. The block threshold — critical only, warning and above, or any finding — and per-rule overrides that re-rank or silence a rule for the whole team. The list covers every SQL rule across engines, searchable and filterable by severity, tier, database and category (see rule ids and categories); a rule name opens its full page in place. The policy is applied by the service to every connected check on Team and above, and never read from a file a commit can change. bolvrk policy pull copies the threshold into bolvrk.json so local runs block where CI blocks.
Team
Tokens — the blv_ team tokens CI uses. Shown once at creation, revoked individually and instantly; the table shows when and from where each was last used, and a new address is an audit event. Members — four roles: admin manages members, policy, connections and the plan; member runs checks and manages connections and tokens; viewer reads everything and changes nothing; billing manages the subscription and nothing else. Invites carry the role the invitee will get and are valid for 7 days. Audit — every team-level action with its actor and time: policy edits, tokens created or revoked, connections and protected databases added or removed, invites, role changes, retention changes, plan changes (Stripe's own recorded with Stripe as the actor). Filter by family, page back as far as it goes. Data — retention and export. Retention is how many days of run log to keep; checks older than that are removed by a daily pass, and the change itself lands in the audit log. Export streams checks, findings and the audit trail as CSV or JSON with the same columns every time — a spreadsheet or a warehouse reads it with no importer.
Billing
The current plan, checkout to Startup, Team or Scale, the Stripe portal for card and plan changes, and the invoice list. Plans are flat per team; a change applies to every token the team already has. Admins and the billing role see this page.
Notifications
Slack, Discord and signed generic webhooks as on the notifications page, with a test button per channel.