Status: Phase 1 build in progress. Beta: Agent Memory + Artifacts
┌──────────────────────────────────────────────────────────────┐
│ PRINCIPAL │
│ Human: Siri / Access-gated board / forwarded mission email │
└───────────────────────┬──────────────────────────────────────┘
▼
┌──────────────────────────────────────────────────────────────┐
│ ORAL — SOLE OPERATOR (api.alfred.report) │
│ Cloudflare Worker + Durable Object + D1 │
│ │
│ Session authority → Mission compiler → Evidence ledger │
│ ↓ ↓ ↓ │
│ Independent gate Budget enforcement Agent Memory │
│ ↓ ↓ ↓ │
│ Voice report (TTS via AI Gateway) │
└──────────────┬───────────────────────────────┬───────────────┘
▼ ▼
┌──────────────────────────┐ ┌────────────────────────────┐
│ FIELD WORKERS │ │ TRUST │
│ Ephemeral, scoped, │ │ D1 event/state record │
│ replaceable │ │ CF Artifacts (source) │
└──────────────────────────┘ └────────────────────────────┘
┌──────────────────────────────────────────────────────────────┐
│ BOARD — THIN EDGE (alfred.report) │
│ Static pages + Access-gated inspection │
│ Service binding to ORAL; no local state │
└──────────────────────────────────────────────────────────────┘
agent-alfred — ORAL operator (missions, sessions, voice, gate)alfred-report — Board (thin edge, static assets, service binding)alfredcustom-mimohermesalfredalfred-commandagent-alfred, alfred-reportX-Tenant-Id header propagationWe have Artifacts repos set up. What's the correct flow for:
Routing MiMo TTS through custom provider. Are we:
Using CF Agent Memory for mission promotion. Are we:
Using DOs for mission state. Are we:
Missions, sessions, events tables. Are we:
Board → ORAL via service binding. Are we:
Cloudflare Access + JWT verification. Are we:
Using wrangler secret put. Is there a better way?
What CF features for:
Should we be using any of these?
Organization: OpenRoyleAl / Iceberg Media
CF Account: 0870b0bdbc14bcd31f43fe5e82c3ee8e
Repository: github.com/OpenRoyleAl/command-os (private)