IAM.Router user manual
Screen-by-screen instructions, exact controls, safe examples, CLI/API alternatives and recovery paths for IAM.Router.
Start with the section that matches the label in the application. Each page explains what the screen does, what to click, which value to enter, how to verify success and how to recover without duplicating an operation.
Documented surfaces
- Dashboard — Summarizes gateway state, task volume, tokens, errors, active clients and average response time.
- History — Finds past routing and inference jobs by request ID, model, status, period and client and opens their source JSON.
- Clients — Registers calling applications, their allowed models and access state for Router.
- Client groups — Groups clients so a shared routing model or quota can be applied.
- API keys — Creates, scopes, rotates and revokes gateway API keys; the full value is shown only at creation time.
- Providers — Shows provider endpoints, protocol/model/perimeter and health used by Router to build permitted routes.
- Routing policy — Defines allowed models, providers, priorities, perimeters and fallback for a tenant or global scope.
- Data classes — Tests text classification and shows sensitivity/data-class evidence before routing policy is applied.
- Webhooks — Delivers signed routing events to an external system with controlled event types and retries.
- Routing audit — Explains why a provider/model/perimeter was selected, which rules passed and which candidates were rejected.
- Prompt injection alerts — Shows detected injection events, severity, the related request and investigation entry point.
- Tenants — Manages organization routing scope, state and baseline constraints.
- Models — Summarizes logical aliases, provider endpoints, context, capabilities and current model health.
- Gateway metrics — Shows Router technical metrics and lets an operator inspect a selected period.
- Usage — Aggregates requests, tokens and cost by tenant, client, model and period.
- Quotas — Sets and shows tenant request, token and budget limits with current consumption.
- Playground — Tests a Route decision or full Chat inference with a model alias without changing production policy.
- Classifier health — Shows classifier throughput, fail-safe/tier-3 rates and latency for a period.
- Live events — Shows a bounded live stream of routing events for operational diagnosis.
- Alert rules — Creates alert conditions for Router metrics, tenant scope and thresholds.
- Workflows (preview) — Preview surface for agent-runner workflows; hidden unless VITE_ROUTER_SHOW_PREVIEW is enabled.
- WASM modules (preview) — Preview surface for agent-runner modules; hidden by a feature flag in standard environments.
- E2EE (preview) — Preview surface for agent-runner E2EE key state; it is not the general Router key-management screen.
Application
How to use DashboardSummarizes gateway state, task volume, tokens, errors, active clients and average response time.
How to use HistoryFinds past routing and inference jobs by request ID, model, status, period and client and opens their source JSON.
How to use ClientsRegisters calling applications, their allowed models and access state for Router.
How to use Client groupsGroups clients so a shared routing model or quota can be applied.
How to use API keysCreates, scopes, rotates and revokes gateway API keys; the full value is shown only at creation time.
How to use ProvidersShows provider endpoints, protocol/model/perimeter and health used by Router to build permitted routes.
How to use Routing policyDefines allowed models, providers, priorities, perimeters and fallback for a tenant or global scope.
How to use Data classesTests text classification and shows sensitivity/data-class evidence before routing policy is applied.
How to use WebhooksDelivers signed routing events to an external system with controlled event types and retries.
How to use Routing auditExplains why a provider/model/perimeter was selected, which rules passed and which candidates were rejected.
How to use Prompt injection alertsShows detected injection events, severity, the related request and investigation entry point.
How to use TenantsManages organization routing scope, state and baseline constraints.
How to use ModelsSummarizes logical aliases, provider endpoints, context, capabilities and current model health.
How to use Gateway metricsShows Router technical metrics and lets an operator inspect a selected period.
How to use UsageAggregates requests, tokens and cost by tenant, client, model and period.
How to use QuotasSets and shows tenant request, token and budget limits with current consumption.
How to use PlaygroundTests a Route decision or full Chat inference with a model alias without changing production policy.
How to use Classifier healthShows classifier throughput, fail-safe/tier-3 rates and latency for a period.
How to use Live eventsShows a bounded live stream of routing events for operational diagnosis.
How to use Alert rulesCreates alert conditions for Router metrics, tenant scope and thresholds.
How to use Workflows (preview)Preview surface for agent-runner workflows; hidden unless VITE_ROUTER_SHOW_PREVIEW is enabled.
How to use WASM modules (preview)Preview surface for agent-runner modules; hidden by a feature flag in standard environments.
How to use E2EE (preview)Preview surface for agent-runner E2EE key state; it is not the general Router key-management screen.
IAM.Router: CLI and APICopy-safe command examples, expected results and recovery rules for IAM.Router.