7.5 KiB
7.5 KiB
CHORUS Services Development Project Plan (v2, Branded)
1. Executive Summary
CHORUS Services is an enterprise-ready distributed AI orchestration platform designed to eliminate context loss, reduce hallucinations, and enable scalable multi-agent collaboration. All naming, interface, and UI/UX conventions have been updated to the CHORUS Services brand and tech-focused visuals, using Ant Design 5+ for the front-end foundation.
2. Updated Technology Stack Overview
- Frontend/UI: React (Next.js 13+), Ant Design 5+ (component library and theming), Framer Motion (animations/parallax), CSS-in-JS with @ant-design/cssinjs
- Backend/API: FastAPI (Python) for APIs; Node.js for certain orchestrator tools
- P2P/Mesh: libp2p (Go or Rust) for BZZZ agent network
- Context Management: SLURP context curator and archive (Python/FastAPI with Postgres/SQLite)
- Distributed Reasoning: HMMM meta-discussion/consensus service
- Feedback/RL: COOEE feedback service (Python/FastAPI, integrated with RL agents)
- Workflow Engine: WHOOSH orchestration engine (integrates natively with BZZZ, HMMM, and SLURP)
- Audit Log: Hypercore event log (Node.js or Rust)
- Networking/Security: Tailscale (P2P overlay), Cloudflare as edge/proxy
- Automation/DevOps: Ansible
- Model Hosting: Ollama (on cloud/VPS), BZZZ container integration
3. Feature/Naming Reference
| Subsystem | Updated Name | Role | Old Name |
|---|---|---|---|
| Orchestrator | WHOOSH | Main workflow/control hub | Hive |
| P2P Comm. | BZZZ | Agent mesh/coordination | Bzzz |
| Reasoning | HMMM | Collaborative pre-task logic | Antennae |
| Context | SLURP | Context curation/storage | HCFS |
| Feedback | COOEE | RL feedback/tuning platform | RL Tuner |
| Log | Hypercore Log | Tamper-evident, audit | Hypercore |
4. Project Modules and Feature Codes
(For traceability: CHORUS-* codes)
| Code | Module | Feature Name | Description |
|---|---|---|---|
| WHOOSH.01 | WHOOSH | AgentRegistry | Register/configure agents; display real-time health/status |
| WHOOSH.02 | WHOOSH | WorkflowEditor | Visual workflow editor (Ant Design UI, React Flow integration) |
| BZZZ.01 | BZZZ | MeshNetworking | Mesh node join, peer discovery (libp2p, Tailscale) |
| BZZZ.02 | BZZZ | P2PTaskFlow | Event and coordination message system |
| HMMM.01 | HMMM | MetaDiscussion | Structured agent reasoning and consensus before action |
| SLURP.01 | SLURP | CuratorEngine | Event classification, context promotion, RL signals |
| SLURP.02 | SLURP | StorageAPI | Context archive, Postgres/SQLite integration |
| COOEE.01 | COOEE | FeedbackCollector | REST API + message bus for feedback ingestion |
| COOEE.02 | COOEE | RLPolicyAdapter | RL-based tuning of promotion/curation rules |
| LOG.01 | Hypercore Log | EventIngestion | Append-only event logging (Node.js/Rust) |
| LOG.02 | Hypercore Log | LogExplorer | UI for browsing/auditing logs (React + Ant Design table/views) |
5. Tech Implementation Notes
Ant Design UI System
- Use Ant Design’s dark mode theme as base, customized for CHORUS colors (#1a1a1a, #007aff, #30d158, #f2f2f7)
- All buttons, cards, tables, and forms should use the Ant Design system for accessibility and consistency
- Use Framer Motion for scroll-based/premier animations (parallax, reveals, metric counter-ups)
- Network diagrams, logs, and workflow flows: combine Ant Design components with (if needed) custom SVG/Canvas
Backend/APIs
- New endpoint structure should use updated nomenclature:
api/chorus/agent,api/chorus/whoosh/, etc. - API docs generated via FastAPI’s native OpenAPI/Swagger
BZZZ Nodes & Tailscale
- All peer discovery and pubsub must use BZZZ and reference BZZZ in code/docs (not Bzzz or any bee analogies)
- Tailscale for securing cross-cloud/cluster comms; internal services should never be internet-facing
Context/RL
- SLURP’s curation logic should be configurable via Ant Design modal/settings UI (YAML upload + real-time tuning)
- COOEE’s feedback features have prominent performance dashboards (Ant Progress, Trend, and Gauge components)
Logging
- All logs and traces to be named CHORUS Event Log or Audit Log in UI/copy
- Include direct links from module UIs to related event log entries for traceability
6. Sprint Board / Waterfall Plan (Phased with New Names)
Phase 1: Infra & Scaffolding
- Initial WHOOSH orchestration deployment (UI + backend structure)
- BZZZ mesh networking POC via Tailscale in local/lab
- SLURP curation engine v0 (basic event filtering)
- LOG module: event ingestion scaffolding
Phase 2: Core Intelligence Flows
- BZZZ agent integration and mesh-wide task routing
- HMMM pre-task consensus/reasoning mechanism (with meta-discussion UI)
- COOEE feedback API and RL signal loop (basic metric view)
- SLURP role-based context curation (UI for manual tuning/testing)
Phase 3: UX/Dashboards
- Ant Design-powered dashboards for:
- Real-time agent status (WHOOSH.01)
- Context curation health and history (SLURP.01)
- Feedback analytics (COOEE.01)
- Log explorer with filters (LOG.02)
- Public landing page: focus on what CHORUS achieves technologically
Phase 4: Integration & Security
- Harden all BZZZ/WHOOSH communications with Tailscale routing
- Cloudflare Zero Trust on all public dashboards/APIs
- OAuth2/JWT integration for RBAC across the UI
Phase 5: Productionization
- CI/CD pipelines with automated tests and deployments
- Automated backup to Backblaze/Scaleway
- Cost/performance optimization (monitoring, log rotation, model/agent scaling)
7. Key Team Roles
- Frontend Developer (Ant Design + React): UI/UX implementation, parallax and animation integration, technical dashboard builds
- Backend/API Developer (Python/FastAPI): REST API, orchestration logic, BZZZ bridge, SLURP curation implementation
- DevOps/Cloud Engineer: Ansible, Tailscale, AWS/Hetzner integration, security hardening, storage automation
- ML/Feedback Specialist: COOEE RL policy, feedback analysis, SLURP rule optimization, trace/data QA
- Documentation/QA Engineer: API/SDK docs, audit storybook, minimization of jargon, clear modular handover
8. Launch Checklist
- All modules rebranded in UI, code, and docs (WHOOSH, BZZZ, HMMM, SLURP, COOEE, CHORUS Event Log)
- Ant Design v5+ and theme applied across UI, with Framer Motion for animation
- Metrics dashboards highlight context retention, hallucination reduction, and cross-agent collaboration rates
- No residual bee/honeycomb/hexagon branding or component naming
- Landing page language and imagery emphasize technological innovation and enterprise readiness
- Automated security, monitoring, and backup pipelines in place