Dashboard

The dashboard is the first page to check after opening CPAMP. It does not replace request details. It answers the first operational questions: is the runtime connected, are requests arriving, did failures increase, did cost jump, and are there account or runtime warnings.

If you already know which account or request is involved, go directly to Monitoring, Auth Files, or Codex Inspection.

What To Check First

• Connection and version: confirm that the panel is connected to CPA or Manager Server and that the runtime mode matches your deployment.
• Quick stats: API keys, auth files, AI providers, and model counts. These tell you whether configuration is being read.
• Requests and cost: today's requests, success rate, average latency, tokens, and estimated cost.
• Collector status: check whether request monitoring is running, whether the queue can be read, and whether recent collection succeeded.
• Health alerts: log errors, connection issues, unavailable monitoring, or missing configuration show up here first.

Dashboard data comes from Manager Server's local SQLite database and the CPA usage queue. Cost is estimated from request events and model prices; it is not a provider invoice.

Common Workflow

1. Start with connection state and collector state.
2. If request count is zero, confirm that clients are actually sending traffic through CPA, then open Request Monitoring Troubleshooting.
3. If success rate dropped, open Monitoring and filter by status, model, account, or API key.
4. If cost increased, open Usage Analytics and break it down by model, account, and caller.
5. If an alert points to logs or runtime state, open Logs and System before changing configuration.

If Dashboard Is Empty

If Dashboard stays empty, do not start by changing filters. Check:

1. Client requests actually pass through CPA.
2. CPA usage publishing is enabled.
3. The CPAMP collector is running.
4. CPA URL and CPA Management Key are correct.
5. Usage queue retention is long enough for events to be collected.

For the full sequence, see Request Monitoring Troubleshooting.

Usage Tips

• If success rate drops, open Monitoring and filter by status code, model, and account.
• If cost spikes, open Usage Analytics and compare model and account breakdowns.
• If one account looks unhealthy, open Auth Files, Quota, or Codex Inspection.
• If login or authorization looks wrong, open OAuth Login, then return to Auth Files.
• If system state looks wrong, open Logs and System and collect version and log context.