Skip to main content
When you log in to AgentVista, the dashboard is your command center for all your agents and infrastructure. It surfaces the most important signals — cost, success rate, recent activity — without requiring you to build queries from scratch.
If you are a new user, you will see an onboarding checklist at the top of the dashboard. It walks you through three steps: generating an API key, installing the SDK, and sending your first trace. The checklist disappears once all three steps are complete.

​
Views

​
Agents list

The agents list at /agents shows every agent that has sent at least one trace. For each agent, the table displays: Click any row to open that agent’s detail view. The status indicator derives from the success rate: 80% or above is healthy, 50–79% is degraded, below 50% is failing, and no outcome signal is inactive.

​
Agent detail

The agent detail view at /agents/{agentId} focuses on a single agent. It contains:
  • Runs over time — a time-series chart of run volume
  • Cost chart — daily cost trend for this agent
  • Success rate chart — success rate over time
  • Recent traces — a list of the most recent traces with status, duration, cost, and outcome
Click any trace in the recent traces list to open the trace detail view.

​
Trace detail

The trace detail view renders a single trace as a waterfall. Each span in the trace is displayed as a row ordered by started_at, with indentation representing the parent-child hierarchy (derived from parent_span_id references). Span types include:
  • agent — the top-level agent execution span
  • llm — an individual LLM call, showing model, input/output tokens, and cost
  • tool — a tool invocation (API call, database query, function call)
  • http — an outbound HTTP request
  • db — a database query
  • custom — any span type you define yourself
Each span row shows its name, status, duration, and — for LLM spans — the model and token counts. Spans with a failed status are highlighted in red.

​
Cost dashboard

The cost dashboard gives you a breakdown of total AI spend across all your agents. You can see:
  • Total spend over a selected time range
  • Spend broken down by agent
  • Spend broken down by model
Use this view to identify which agents are driving the most cost and whether a model change is having a measurable impact.

​
Service map

The service map auto-discovers the topology of all your services and agents from ingested trace data. It requires no manual configuration. See the Service map page for full details.

​
Metrics explorer

The metrics explorer lets you query and visualize metric streams that have been ingested via OTLP or the AgentVista SDK. You can select a metric, apply filters, and view it as a time-series chart. The log search view gives you full-text search across all structured log records. You can filter by service name, severity, and time range, and click through to the trace a log record belongs to.

​
Alerts

The alerts view shows all your active alert configurations — what type each alert is, the threshold, the delivery channel, and when it last fired. See the Alerts page for details on creating and managing alerts.

​
Configurable dashboard panels

The main dashboard at /dashboard is configurable. You can:
  • Add, remove, and rearrange panels
  • Choose from a template gallery of pre-built panel layouts for common stacks
  • Create multiple named dashboards and switch between them using the dashboard selector
To apply a template, click Templates in the tab bar, select a template, and click Apply. A new dashboard is created with pre-configured panels and selected automatically. If you have more than one dashboard, a dropdown selector appears in the top-right corner of the dashboard tab so you can switch between them.