Spanda

Attention Engine

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

Purpose

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.

Responsibilities

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

Entity integration

Field Description
Entity.attention EntityAttentionSnapshottop_priority, queue_depth, focused_event, suppressed_count

Populated during enrich_entity_autonomy() from health/readiness severity mapping.

Language

attention_policy MissionFocus {
    rule suppress_low_priority;
    rule boost_critical_health;
}

Types (spanda-autonomy)

AttentionPolicy, AttentionScore, EventPriority, SignalPriority, AttentionWindow, SuppressionRule

Examples

CLI

spanda alerts analyze
spanda alerts fatigue-report

API

Surface Endpoint / RPC
REST GET /v1/autonomy/attention
gRPC GetAutonomyAttention
SDK AttentionClient::queue()
Entity GET /v1/entities/{id}/autonomyattention field

Control Center

Attention Queue section in the Cognitive & Resilience tab.

Integrations

See responsibility-matrix.md.