Ecosystem architecture

The technical system is a governed tool network around signed road alerts.

Looma should be presented as the road-signal product. LunaOS.ai, MCP, OpenSyber.cloud, ClawPipe.ai, Amliq.finance, and LunaForge become the operating system around it: runtime, tools, governance, AI routing, compliance, and code-risk evidence.

LoomaProduct network

Signed browser identity, road-alert payloads, edge relay, nearby feed, and operator console.

Fleet/OEM device adapters, tenant workspaces, structured event schemas, and integration SLAs.

LunaOS.aiAgent runtime

Target platform for AI-native backend operations and internal product workflows.

Move pilot tenants, agent state, operational APIs, and deployment workflows into a managed runtime.

MCPTool interface

Integration pattern, not yet a public Looma tool registry.

Expose typed tools for alerts, evidence, incident reports, partner screening, cost routing, and audit.

OpenSyber.cloudGovernance

Ecosystem governance platform with MCP control, agent monitoring, policy, and audit concepts.

Require approvals, scopes, logs, and policy checks around every Looma agent action.

ClawPipe.aiAI routing

Ecosystem AI pipeline for model routing, cache, provider choice, budgets, and tracing.

Route Looma summaries, pilot reports, and compliance prompts through controlled model pipelines.

Amliq.financeCompliance intelligence

AML and counterparty intelligence platform for screening, cases, reports, and audit chains.

Screen fleet customers, vendors, resellers, procurement entities, and higher-risk counterparties.

LunaForgeEngineering evidence

Sibling code-risk and context engine with MCP-oriented evidence export.

Generate change-impact evidence whenever safety-message logic, relay code, or agent tools change.

MCP surface

Use MCP to make the product programmable without giving agents unlimited access.

MCP should be the typed interface between Looma's product network and internal agents. The point is not novelty. The point is controllable capabilities, audit, and repeatable integrations.

Tool
Actor
Permission
Business value
looma.alerts.nearbyTriage AgentRead

Fetch recent signed alerts near a route, vehicle, depot, or pilot zone.

looma.alerts.submitSignedVehicle AdapterWrite with signature

Submit a structured hazard event with identity, signature, location, and expiration metadata.

looma.incidents.summarizeBriefing AgentRead plus generate

Cluster events and produce a plain-language incident brief for an operator.

looma.evidence.exportPilotReportPilot Report AgentApproval required

Create investor and customer evidence from relay activity, uptime, message quality, and response patterns.

opensyber.auditAgentRunGovernance AgentRequired control

Record the user, tool call, input scope, approval, output, and policy result for regulated actions.

clawpipe.routePromptAI RouterSystem service

Choose model, budget, cache policy, and trace behavior for generative tasks.

amliq.screenCounterpartyCompliance AgentApproval required

Run KYB/KYC-style checks on pilot entities, vendors, channel partners, and procurement contacts.

lunaforge.analyzeChangeImpactEngineering AgentRead repository

Map dependency impact and produce evidence before code touching relay, identity, or safety schema ships.

Safety events

Alert type, location, timestamp, signature, sender identity, expiration, and confidence metadata.

Short retention by default. Minimize personally identifiable data.

Pilot evidence

Message volume, delivery quality, uptime, route coverage, alert clusters, and response notes.

Usable for investors and customers after redaction and tenant approval.

Compliance records

Counterparty screening results, case notes, report exports, and audit trail references.

Handled through Amliq-style controls with access separation.

Agent traces

Prompt inputs, model route, tool calls, approvals, policies, outputs, and reviewer actions.

Governed through OpenSyber-style logging and retention policy.

Integration path

Sequence the ecosystem work so it raises credibility at each step.

Phase 1

Expose Looma tools

Turn alert fetch, signed submit, incident summary, and evidence export into internal MCP tools. Route non-critical generation through ClawPipe.

Phase 2

Govern every agent run

Attach OpenSyber policies to tool calls. Add approval gates for writes, exports, compliance checks, and external customer communications.

Phase 3

Operationalize pilots

Use LunaOS.ai for tenant state, app operations, APIs, and backend agents. Use Amliq for customer and partner screening.

Phase 4

Move toward production roads

Add OEM/fleet integrations, hardware partners, C-V2X/SCMS planning, security review, legal review, and certification milestones.