Global Awareness Engine: Unified Cross-Signal Intelligence

An octopus processes information from eight arms simultaneously — each arm sensing independently, yet contributing to a unified model of the environment. The Global Awareness Engine gives OctopusOS the same capability: a unified perception layer that consolidates all signal categories into a single awareness plane, enabling cross-signal correlation, influencer tracking, meme propagation analysis, and competitive positioning.

1. Why Unified Awareness Matters

OctopusOS already has deep subsystem-level awareness: Self-Awareness monitors internal health, Resource Governor watches the environment, ScreenOS tracks workspace activity, Internet Awareness monitors online mentions, and the Execution layer tracks human feedback. But each subsystem operates in its own silo.

The critical intelligence gap is cross-signal correlation: when a CPU spike (environment) coincides with a spike in negative mentions (internet) and increased task failures (human feedback), each subsystem reports its own anomaly — but no one connects the dots.

The Global Awareness Engine (GAE) sits above all subsystems as a read-only consumer, normalizing their outputs into a unified GlobalSignalEvent schema and performing cross-category analysis that no individual subsystem can achieve.

2. Five Perception Categories

Every signal in the OctopusOS universe is classified into one of five fundamental awareness categories. These categories map directly to existing subsystems while providing a unified vocabulary for cross-signal reasoning.

3. The GlobalSignalEvent: A Universal Perception Schema

At the core of GAE is the GlobalSignalEvent — a frozen, immutable dataclass that provides a unified view of any event from any subsystem.

@dataclass(frozen=True)
class GlobalSignalEvent:
    event_id:        str                  # Unique identifier
    category:        AwarenessCategory    # internal | environment | workspace | internet | human
    source_kind:     str                  # Origin subsystem (self_awareness, resource_governor, etc.)
    ts_ms:           int                  # Millisecond timestamp
    summary:         str                  # Human-readable description
    severity:        EventSeverity        # info | warning | critical
    payload:         dict[str, Any]       # Structured event data
    signal_id:       str                  # Link to SignalOS SignalEnvelope
    correlation_id:  str                  # Cross-signal correlation chain
    metadata:        dict[str, Any]       # Extensible metadata

Design principles:

• Category-first — every event is tagged with one of five categories, enabling cross-category queries
• Severity-aware — automatic severity inference from source data (e.g., CPU > 90% = critical)
• SignalOS integration — signal_id links to SignalOS SignalEnvelope for full causal tracing
• Subsystem-agnostic — normalizers handle the translation from subsystem-specific formats

4. Signal Normalization Pipeline

GAE includes six normalizers that translate subsystem outputs into GlobalSignalEvent instances. Each normalizer is a pure function with no IO dependencies.

5. Influencer Detection

The Influencer Detection module identifies high-impact authors across platforms and monitors their activity. This goes beyond simple mention counting — it uses a weighted composite scoring formula that considers multiple dimensions of influence.

Influence Score Formula

The composite influence score (0–100) is computed using four weighted factors:

score = (
    0.25 × min(100, log2(1 + mentions) × 15)
  + 0.30 × min(100, log2(1 + engagement) × 10)
  + 0.15 × min(100, abs(sentiment) × 100)
  + 0.30 × min(100, platforms × 25)
)

Influencer Alerts

When an influencer exceeding the configured threshold (default: 50.0) publishes new content, GAE generates an InfluencerAlert with the author’s profile, the triggering mention, and a human-readable reason. These alerts are emitted as high-priority SignalEnvelope messages into SignalOS.

6. Meme Propagation Graph

Content doesn’t stay on one platform. A Hacker News post gets shared on Reddit, tweeted about, and forked on GitHub. The Meme Propagation module tracks how content spreads across platforms by building directed propagation graphs.

Three-Tier Matching

Propagation Metrics

Each MemeGraph captures:

• Origin — the earliest mention, identified as the propagation source
• Propagation velocity — platforms reached per hour
• Total reach — sum of engagement scores across all nodes
• Platform count — number of distinct platforms the content spread to

7. Competitive Positioning

The Competitive Positioning module provides automated share-of-voice analysis, sentiment benchmarking, and trend divergence detection — the kind of competitive intelligence that typically requires dedicated market research tools.

Share of Voice (SOV)

For each topic (brand, product, competitor), GAE computes:

• Mention share — what percentage of total mentions belong to this topic
• Engagement share — weighted by platform engagement scores
• Sentiment average — mean sentiment across all mentions
• Trend direction — rising, stable, or declining

Sentiment Gap Analysis

The sentiment gap between your topic and each competitor is computed as a simple delta: primary_sentiment - competitor_sentiment. Positive gaps indicate sentiment advantage; negative gaps signal competitive vulnerability.

Trend Divergence Detection

When a competitor’s trend direction diverges from your own, GAE generates labeled warnings:

8. Global Awareness Snapshot

The awareness snapshot is the crown jewel of GAE — a periodic, holistic summary that correlates signals across all five categories and generates actionable insights.

Cross-Signal Anomaly Detection

GAE detects four categories of cross-signal anomalies:

1. Environment + Internal stress — Resource pressure coinciding with internal issues suggests infrastructure-driven failures
2. Critical signal concentration — Many critical-severity events within a window indicate a developing incident
3. Internet spike — Sudden increase in internet mentions (>10 in a window) may indicate viral content or PR event
4. Human feedback failures — Elevated failure rate in execution outcomes signals workflow issues

Temporal Correlation

Events are bucketed into hourly windows and analyzed for co-occurrence. When multiple categories spike simultaneously, GAE reports the correlation — for example: “Environment and Internal events co-occurred in 3 hourly windows, suggesting infrastructure-driven internal issues.”

9. SignalOS Integration

GAE produces four types of SignalEnvelope messages that feed directly into the SignalOS bus:

This integration means GAE insights flow through SignalOS governance (risk grading, routing, replay) and eventually into the Flywheel learning loop — creating a closed intelligence cycle.

10. Architecture Overview

Layer Placement Rules

• L1 (Contracts + Port) — kernel/contracts/global_awareness.py and kernel/ports/global_awareness/interfaces.py. Pure type definitions and protocol.
• L2 (Domain Logic) — kernel/domains/global_awareness/. Six pure-function modules with zero IO.
• L3 (Runtime) — kernel/runtime/_wl_global_awareness.py. Worker Loop Mixin scheduling four periodic tasks.
• L4 (API + Dashboard) — server/shared/adapters/http/_routes_global_awareness.py. REST API and HTML dashboard.

11. API Surface

GAE exposes twelve REST endpoints:

12. Implementation Metrics

13. Design Philosophy

GAE embodies three architectural principles:

1. Read-only consumption. GAE never modifies existing subsystems. It reads their outputs through normalizers, correlates across categories, and emits new signals — all without side effects on the source subsystems. This ensures that adding GAE cannot break any existing functionality.

2. Cross-signal intelligence over depth. Individual subsystems already provide deep domain-specific analysis. GAE’s value is in the connections between subsystems: detecting that a resource pressure spike (environment) correlates with mention sentiment decline (internet) and increased task failures (human). No single subsystem can see this pattern.

3. Progressive enrichment. GAE builds intelligence in layers: raw normalization, then cross-signal anomaly detection, then competitive positioning, then actionable recommendations. Each layer can operate independently, and each adds value even if the layers above it are disabled.

14. Application Scenarios

Real-Time Brand Intelligence

When a mention of OctopusOS goes viral on Hacker News (internet category), GAE tracks the propagation to Reddit, Twitter, and GitHub. The influencer detection module identifies which authors are driving the conversation, and the competitive positioning module shows how this compares to competitor mention patterns — all updated in real-time through the dashboard.

Incident Correlation

During a production incident, GAE correlates environment signals (CPU spike, disk pressure) with internal signals (task queue growth, learning rate drop) and human signals (increased failure rate) to build a unified incident timeline. The cross-signal anomaly detector flags the correlation before any single subsystem would escalate.

Competitive Early Warning

When a competitor’s mention velocity suddenly increases while yours stays flat, GAE’s trend divergence detector generates a WATCH or RISK alert. This early warning — powered by share-of-voice analysis across all monitored platforms — gives the team time to respond before the competitive gap widens.

Content Strategy Feedback Loop

GAE’s meme propagation graphs reveal which content resonates across platforms and which stays contained. By correlating propagation velocity with influencer engagement, the engine identifies which authors and platforms are most effective distribution channels — actionable intelligence for content strategy.