Scout is the control plane for production agents: prompt guard, approvals, live monitoring, kill switch, and audit exports.
A governance control plane: audit every agent action, enforce policy before execution, halt compromised runs, and generate board-ready compliance reports.
See active agent sessions, event volume, and risk score in real time. Posture banner shows SAFE, AT RISK, or BREACH at a glance.
Three-layer defence scans user input, tool responses, and model context before every action. Regex pre-filter, LLM judge, and output guard block malicious instructions in real time.
When policy returns require_approval, the agent pauses and waits. Operators approve or reject from the dashboard before execution continues.
Live event stream, anomaly charts, and kill switch controls for security engineers.
Scout automatically maps your runtime data to EU AI Act risk tiers (Minimal, Limited, High Risk) and UK DSIT alignment levels. Gaps listed with article references.
A separate view for CISOs, compliance officers, and boards. Shows framework readiness scores (EU AI Act, GDPR, SOC2, ISO 42001) and exports a one-click PDF board report.
Sector-specific governance packs for UK property (AML 2017, RICS), legal (SRA), and finance (FCA Consumer Duty). Maps agent events to named regulations and articles.
Critical notifications routed to Telegram, email, Slack, or Discord when incidents occur or approvals are requested.
Halt all governed agents in one action from the dashboard. Human control over autonomous AI. Arms automatically; fires when you say so.
Scout records every agent action with a full audit trail. Optional on-chain attestations via ERC-8004 identity and ERC-8126 verification provide tamper-evident proof when wallet configuration is active. Live metrics at carinaai.uk/trust.
Tool calls, policy decisions, session events, and operator approvals all captured with timestamps. Exportable as JSON, CSV, or PDF.
ERC-8004 agent identity and ERC-8126 WAV attestations create a public record of governance posture when wallet-backed verification is enabled.
The governance dashboard generates a PDF covering EU AI Act tier, framework readiness scores, attestation history, and recommended actions.
Carina pairs in setup: run carina setup with credentials from your provisioning email. Other agents use the HTTP events API or SDKs. Public read-only metrics: carinaai.uk/trust.
Carina is the reference workforce agent and pairs in one setup wizard. Hermes, AutoGPT, OpenClaw, Oddyseus, and custom stacks connect via HTTP events or SDK adapters.
Carina product details (tools, packs, channels) live on carinaai.uk.
One Scout licence per deployment. Monitor gives visibility; Pro adds response controls and alerts. Carina Core is free on carinaai.uk.
Console URL and API key emailed at provisioning.
Deploy Scout beside your agent. Carina pairs in setup; other stacks use proxy mode, the HTTP events API, or SDKs.
Run Scout on your VPS. Your console URL and API key come from the provisioning email.
SCOUT_API_KEY=your_key npm start
Node.js agents use the SDK or HTTP events API. Carina pairs automatically with carina setup.
npm install @verlox/labyrinth-scout
Select an industry template or a generic policy pack. Scout starts classifying events against named regulations immediately.
POST /api/governance/templates/uk-property/apply
Scout Pro adds regulatory classification, audit trails, templates, approvals, and a kill switch for £99/mo per deployment.