The app dashboard at app.openlegion.ai is where you monitor your instance status and manage your subscription. For agent configuration and fleet management, use the engine dashboard on your subdomain (covered at the bottom of this page).
Fleet status
When your instance is running, the dashboard shows:
- Status badge — Online, Starting up, Resizing, Migrating, or Paused
- Subdomain — your fleet URL (
{subdomain}.engine.openlegion.ai) with a copy button
- Plan summary — current plan, agents, browsers, projects, and credit balance
- Open Dashboard button — the primary CTA; single-sign-on into your engine
The page auto-refreshes every 10s in transitional states (starting up, resizing).
Migrating is a defined status that the UI renders but is not reachable from current code paths — it’s staged for a future feature.
Open Dashboard
This button is the only supported way to enter your engine from the app. The app generates a short-lived HMAC-SHA256 token (5-minute TTL, one-time use), 302-redirects you through the auth-gate sidecar on your VPS, which then sets a 24h ol_session cookie. Direct navigation to your subdomain (e.g., from a bookmark) requires an existing session cookie. If your cookie has expired, return to the app and click Open Dashboard again.
Notifications and banners
The fleet dashboard surfaces banners for: past-due payment, cancelled subscription (countdown), paused instance, resizing, migrating, and unhealthy.
Instance states
The fleet view condenses 15 internal status values into the user-facing badges above. The most common transitional and failure states you’ll encounter:
| State | Meaning |
|---|
Online (running) | Your fleet is running normally |
Starting up (queued → provisioning → bootstrapping → awaiting_configuration → configuring) | Initial provisioning or post-restart warm-up — auto-refreshes |
| Resizing | Server is being resized for a plan change (~30-60s downtime) |
| Paused | Instance paused due to a past-due payment (resumes automatically when payment succeeds) |
| Unhealthy | Engine not responding to health checks. Auto-restart attempts in progress (max 3). |
Provisioning Failed (failed / bootstrap_failed) | Server setup didn’t complete — retry runs from scratch (VPS is cleaned up first) |
Configuration Failed (configuration_failed) | Server is healthy but configuration didn’t complete — retry just re-runs /configure (VPS is kept) |
Decommissioned (deprovisioned) | Subscription ended or unrecoverable failure — the app shows a Decommissioned card with a new-subscription CTA |
Settings
The Settings page shows your current plan and lets you:
- Change plan — select a different tier or switch between monthly and yearly billing
- Resume — reactivate a cancelled subscription before the period ends
- Cancel — cancel your subscription (access kept until end of billing period)
- Keep current plan — cancel a pending downgrade so you stay on your current tier at renewal
See Plans & Billing for details on upgrades, downgrades, and payment handling.
Credits
The Credits page (/credits in the app) shows your current OpenLegion credit balance and lets you top up. Every managed instance ships with the openlegion/openai/gpt-5.4 credit-backed default model; welcome credits are granted on first subscription activation.
What the app dashboard does not do
The app dashboard is intentionally narrow:
- No user-initiated pause / resume (pause only triggers after 3 days past-due)
- No region picker (server location chosen by provisioner fallback chain)
- No “rotate access token” UI
- No re-configure API keys UI (manage those from the engine’s Settings → API Keys tab)
- No multi-instance support — one VPS per subscription
- No log streaming, per-agent metrics, or backup/restore UX
- No direct agent control, chat UI, or project management — all of that lives on the engine dashboard
Engine dashboard
The engine dashboard at {subdomain}.engine.openlegion.ai is where you do the actual work with your agents. Open it via the app’s Open Dashboard button (see SSO note above).
The engine SPA is built on Alpine.js + Tailwind with no build step. It serves 143 dashboard API endpoints and streams real-time updates over a /ws/events WebSocket (50 event types, 500-event ring buffer).
Four top-nav tabs
| Tab | Internal id | What you do here |
|---|
| Chat | chat | Multi-agent chat. Your primary conversation surface. Per-token streaming chat over a dedicated endpoint. |
| Work | workplace | Kanban (Pending / Working / Blocked / Done), “Needs You” panel, activity feed with pinned blockers, recently delivered previews. |
| Team | fleet | Agent grid — inspect, edit, and manage individual agents (model, budget, soul, instructions, permissions). |
| Settings | system | 11 sub-tabs (below). |
Settings sub-tabs (11)
Activity, Costs, Automation, Integrations, API Keys, Wallet, Network, Storage, Operator, Browser, Settings.
Highlights:
- Costs — per-agent / per-project LLM cost ledger with daily and monthly budgets.
- Automation — cron jobs and heartbeats (5-field cron or
every N[s/m/h/d] interval).
- Integrations — channel setup (Telegram / Discord / Slack / WhatsApp / Webhook) and webhook creation (webhooks are dashboard-only).
- API Keys — add or replace LLM provider keys (BYOK overlay on top of the OpenLegion credit proxy).
- Wallet — seed reveal is once-only (HTTP 410 thereafter); per-agent EVM + Solana addresses; spend caps and rate limits.
- Browser — Camoufox flags, CAPTCHA solver config, device profiles, fingerprint controls. (Fingerprint burn detection does not auto-rotate — operator action required.)
- Operator — operator-only control plane: edit agents, project management, pending actions.
Browser viewer
Each agent gets its own Camoufox (stealth Firefox) instance in the shared browser-service container, lazy-spawned on first use. The dashboard provides a per-agent live VNC view via the engine’s /agent-vnc/{agent_id}/{path} proxy. The VNC proxy only accepts the ol_session cookie — agent Bearer tokens are rejected. 