Functional domain: Attention Engine
Status: Beta — health-mapped event prioritization for Control Center, telemetry, and API
surfaces.
Canonical architecture: cognitive-resilience-architecture.md Prior short guide: attention-system.md
Reduce event overload and alert fatigue by prioritizing signals that matter for safety, missions, and operators. The Attention Engine ranks, suppresses, aggregates, and focuses events — it does not replace alerting infrastructure; it governs what surfaces first.
| Responsibility | Mechanism |
|---|---|
| Prioritize critical events | EventPriority::Critical + severity boost |
| Boost mission events | Attention policy rules in .sd |
| Suppress routine noise | Habituation policy (spanda-autonomy::habituation) |
| Focus operator attention | Ranked AttentionWindow |
| Field | Description |
|---|---|
Entity.attention |
EntityAttentionSnapshot — top_priority, queue_depth, focused_event, suppressed_count |
Populated during enrich_entity_autonomy() from health/readiness severity mapping.
attention_policy MissionFocus {
rule suppress_low_priority;
rule boost_critical_health;
}
spanda-autonomy)AttentionPolicy, AttentionScore, EventPriority, SignalPriority, AttentionWindow,
SuppressionRule
spanda alerts analyze
spanda alerts fatigue-report
| Surface | Endpoint / RPC |
|---|---|
| REST | GET /v1/autonomy/attention |
| gRPC | GetAutonomyAttention |
| SDK | AttentionClient::queue() |
| Entity | GET /v1/entities/{id}/autonomy → attention field |
Attention Queue section in the Cognitive & Resilience tab.