Platform prioritizes deterministic evaluation, validator integrity, and minimal trust assumptions. Security controls are enforced at the network, consensus, and runtime layers.

1. Stake-weighted validator set: validators must meet minimum stake requirements.
2. Signed P2P messages: every submission, evaluation, and vote is signed with the validator hotkey.
3. PBFT-style consensus: final state is accepted only with >= 2f + 1 approvals.
4. WASM sandbox: challenge execution is isolated with strict runtime policies.
5. Auditability: state changes and challenge updates are anchored to chain epochs.

• Byzantine validators: mitigated by quorum thresholds and stake weighting.
• Sybil attempts: minimum stake requirements and metagraph verification.
• Challenge tampering: challenge metadata is signed and consensus-approved.
• Resource exhaustion: WASM runtime enforces CPU, memory, and I/O caps.

flowchart TB
    subgraph Network
        P2P[libp2p Mesh]
        DHT[DHT]
        P2P --> DHT
    end

    subgraph Validator
        Signed[Signed Messages]
        Consensus[PBFT Consensus]
        Runtime[WASM Runtime]
        Audit[Audit Logs]
    end

    P2P --> Signed
    Signed --> Consensus
    Consensus --> Runtime
    Runtime --> Audit

flowchart LR
    Validator[Validator Node] --> Runtime[WASM Runtime]
    Runtime --> Policy[Runtime Policy + Limits]
    Runtime --> HostFns[Whitelisted Host Functions]
    Runtime --> Audit[Audit Logs]
    Policy --> Runtime
    HostFns --> Runtime
    Runtime -->|Deterministic outputs| Validator

Docker-backed evaluation is restricted to test environments and guarded by an explicit security policy. Production validators run the WASM runtime only.

flowchart LR
    Tester[Test Harness] --> Broker[Secure Container Broker]
    Broker --> Policy[Security Policy]
    Policy --> Runtime[Container Runtime]
    Runtime --> Logs[Audit Logs]
    Policy --> Broker

• Key management: secrets via env vars or secret managers.
• Network controls: firewall rules limit ingress to required ports.
• Monitoring: health checks and log monitoring detect consensus drift.

• Architecture
• Validator Operations
• Challenges