AppSite Cyber Check

Beta
10 of 16 claims live

Agents

agents.appsite.ca

This app is in beta: it’s live and ready to use, and some of its services are still being completed. Below is every claim on its Home and where it stands.

This record checked an older build. It checked 1.4.423 (Oct 20, 2026, 8:34 a.m.); agents.appsite.ca now serves 1.4.423 (7dc2451). The claims and checks below are about the older build until the next scan.

verify — agents · 1.4.423 001bfd4

What the Home says, checked against the app — 10 backed · 3 partly · 1 not yet · 2 not verified

  • Backed by the appEach runner in the bar names the Claude account that answers its jobs and shows whether it is on, waiting for a sign-in, or off.
  • Backed by the appThe bar lists every job a runner has queued, running or just finished, with its last line and elapsed run time, and each job opens its Cyber Check record.
  • Backed by the appRunners handle five kinds of job: fixes, repairs, scans, deploys and ships.
  • Backed by the appAPI-answered calls are charged in US$, calls answered on the Claude plan via Mac or Cloud Shell cost nothing, and the bar shows their API-equivalent cost.
  • Backed by the appwas partly backedHumans and AI agents sit on one roster, edited inline, with what each may touch.Repaired — the code was fixed and merged: Add a per-agent 'May touch' scope list to every roster record, edited inline and shown on Control Panel. PR ↗ · diff ↗ Re-checked 2026-10-20 08:34 UTC, after the fix: backed by the app ✓.
  • Backed by the appwas partly backedAssignment rules route each trigger's work to the right agent, e.g. a new lead or a failed scan.Repaired — the code was fixed and merged: New assignment router applies rules to new leads/bookings/orders/emails and failed scans, logs it. PR ↗ · diff ↗ Re-checked 2026-10-20 08:34 UTC, after the fix: backed by the app ✓.
  • Backed by the appwas partly backedAny agent can be run on demand from its row on the roster.Repaired — the code was fixed and merged: Generic runAgentOnDemand function + per-row Run now for every roster agent. PR ↗ · diff ↗ Re-checked 2026-10-20 08:34 UTC, after the fix: backed by the app ✓.
  • Partly liveAn activity log records every run and every assignment, with what each run did.Why: The log is displayed, but no code shown writes runs or assignments into it; runAgentNow and the rules write nothing there.
  • On the wayThe platform's own agents (Storyboard, Device ID, Site Consistency and more) are listed beside the tenant's agents.Repaired — the code was fixed, the PR awaits review: [object Object]. PR ↗ · diff ↗ The next scan of the live build re-checks it.
  • Partly liveEach AI agent runs on AppSite's provider configuration or on the tenant's own API keys.Why: The AppSite provider path works, but there is no way to enter or use a tenant's own keys.
  • Partly liveAgents is the registry every AI tool on the platform reports to.Why: Only agents calling through agentAiGenerate must be registered; the runner CLIs, flowRunner AI steps and other gateway callers do not report to Agents.
  • Backed by the appTenants' own AI agents run on the tokens included in their AppSite plan.
  • Backed by the appThe module uses the same tenant identity, permissions and shared records as the rest of AppSite.
  • Backed by the appAgents can be opened in a Desk pane alongside a second module.
  • Not checked yetCourses and the Developer Certificate are included free but only with the Enterprise or Creator plan, opening in Training.Why: Content, not code: plan and course offering.
  • Not checked yetAgents is in the recommended module set for 2 of 19 industries (Wellness, Spiritual).Why: Content, not code: industry recommendation counts.

Checked Oct 20, 2026, 8:34 a.m. against build 1.4.423 (001bfd4). Cyber Check reads the Home a visitor sees, then compares every claim on it with the code of the build that is serving it — a label or a screenshot is not evidence; working code is. It also checks every link, HTTPS, shipped secrets, DNS, the public page, and the registry — then the other way: what the app offers that its Home never says. After a full first pass, it rechecks each app on any night its code, its build or its Home changed.