> For the complete documentation index, see [llms.txt](https://candora.gitbook.io/whitepaper/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://candora.gitbook.io/whitepaper/candora-grid/cortex.md).

# Cortex

Computational Intelligence Infrastructure Layer, Adaptive Orchestration Runtime, and Replayable Intelligence Execution Layer

### Computational Intelligence, Analytics, and Orchestration Layer

Candora Cortex is the computational intelligence layer of Candora Grid. It is responsible for analytics, behavioral modeling, simulation, automation, orchestration, and intelligence-driven interpretation of synchronized system state across the exchange infrastructure.

Cortex serves as the analytical layer of the Grid. It consumes telemetry produced by Orbit, Pulse, Oracle, Risk Engine, Vault, and Connect and transforms that information into structured intelligence used for monitoring, forecasting, operational awareness, automation, and system analysis.

Unlike execution, pricing, risk, and settlement systems, Cortex possesses no financial authority. It cannot modify execution outcomes, alter valuation state, override solvency decisions, authorize settlements, or influence custody operations. Instead, Cortex operates as a bounded computational environment that exists alongside the core infrastructure while remaining independent from it.

This separation preserves a clear distinction between authoritative financial systems and computational intelligence systems. Orbit governs execution, Oracle governs valuation, Risk Engine governs solvency, Vault governs financial state, Pulse governs market-state propagation, and Connect governs external access. Cortex exists to analyze and interpret the behavior of those systems without becoming part of their authority structure.

### Telemetry-Driven Intelligence Model

All Cortex computation originates from synchronized infrastructure telemetry generated throughout Candora Grid. Cortex consumes market-state information from Pulse, execution-state transitions from Orbit, valuation updates from Oracle, solvency information from Risk Engine, financial-state data from Vault, and external interaction telemetry from Connect.

At a system level, intelligence outputs are derived from coordinated infrastructure state:

```
I(t) = f(Pulse, Orbit, Oracle, Risk, Vault, Connect)
```

Because all inputs originate from replayable system-of-record event streams, Cortex remains aligned with authoritative platform state. It does not generate independent market truth, financial state, or risk determinations. Every analytical output remains traceable to underlying infrastructure telemetry.

### Runtime Architecture

Cortex operates as an event-driven computational runtime composed of isolated execution environments designed for analytical and orchestration workloads. These workloads may include behavioral analysis, market analytics, anomaly detection, simulation, automation workflows, strategy coordination, forecasting, and operational intelligence.

Computations execute asynchronously from the exchange's execution-critical infrastructure. Cortex consumes events and produces intelligence artifacts, but it does not directly modify execution, pricing, solvency, settlement, or custody state.

The general processing model follows:

```
System Telemetry → Cortex Computation → Intelligence Outputs → Authorized Consumers
```

Downstream systems may consume selected outputs where permitted by their operating models, but any resulting action remains governed by the receiving system's own rules and authority boundaries.

### Authority Boundary

Cortex is intentionally constrained to preserve execution neutrality and infrastructure integrity. It cannot match orders, determine execution priority, set trade prices, modify valuation state, define margin requirements, initiate liquidations, authorize settlements, alter custody records, or override infrastructure controls.

These responsibilities remain exclusively within Orbit, Oracle, Risk Engine, Vault, Pulse, and Connect.

Cortex may generate analytical observations, forecasts, classifications, simulations, behavioral assessments, and operational recommendations. However, these outputs remain advisory in nature. Final authority always resides with the destination system consuming the information.

This design allows Candora Grid to benefit from advanced analytical capabilities while preserving deterministic infrastructure behavior and clearly defined system boundaries.

### System Coordination

By consuming telemetry from all major infrastructure domains, Cortex maintains a unified analytical view of exchange activity. This enables cross-system analysis that would not be visible from any single subsystem in isolation.

Cortex may be used to analyze market behavior, evaluate execution quality, identify operational anomalies, model liquidity conditions, monitor infrastructure performance, coordinate automation workflows, and simulate potential future scenarios. Because it operates on synchronized infrastructure state, its analysis remains consistent with the broader operating conditions of the exchange.

The objective is not to control infrastructure behavior but to improve visibility, interpretation, and understanding of that behavior.

### Resource Isolation and Scheduling

Cortex operates independently from execution-critical infrastructure and executes all workloads within bounded resource limits. Analytical computation is intentionally isolated from matching, pricing, solvency, settlement, and custody systems to ensure that intelligence workloads cannot interfere with core exchange operations.

When computational demand increases, Cortex may defer, throttle, or reschedule lower-priority workloads according to predefined scheduling policies. This allows analytical services to scale independently while preserving the stability and performance of the underlying exchange infrastructure.

### Shared Stress-State Integration

Cortex consumes the shared stress state defined within Candora's [Operational Model](broken://spaces/MEFZMk8235RQB0qt6s7o/pages/j1jmIWNhXl9E3sWbp0Vx) and may incorporate it into analytical workflows, market-regime classification, forecasting models, simulation environments, and operational monitoring systems.

Cortex may also generate analytical telemetry that contributes to broader infrastructure awareness. Such telemetry remains informational and advisory. It does not independently alter platform behavior, activate controls, or override rule-based infrastructure decisions.

Operational responses to changing stress conditions remain the responsibility of the systems that consume the shared stress state.

### Replay and Deterministic Reconstruction

All Cortex computations are derived from replayable event streams and deterministic computational logic. Given identical telemetry inputs and computational rules, Cortex produces identical analytical outputs.

Reconstruction follows:

```
I_recovered = I_snapshot + Σ Δsystem_events
```

This allows intelligence state to be reproduced across recovery environments, simulation systems, audit workflows, distributed replay environments, and historical analysis platforms.

Replayability ensures that analytical conclusions remain verifiable, reproducible, and consistent with the underlying system behavior from which they were derived.

### System Role Summary

Candora Cortex serves as the computational intelligence layer of Candora Grid. It provides analytics, modeling, simulation, orchestration, automation, forecasting, and behavioral analysis while remaining fully separated from execution, pricing, solvency, settlement, and custody authority.

All intelligence generation, analytical computation, simulation, and orchestration within Candora Grid ultimately originate from Cortex, making it the foundational computational and analytical layer of the exchange infrastructure.
