Jinban.app

System Architecture

The platform is built as a privacy‑first, EU‑hosted SaaS with optional self‑hosted deployment. All data stays within OVH data centers (France/Canada).

flowchart LR
    subgraph Frontend["Frontend (Stimulus.js + LiveView)"]
        UI["UI Components<br/>(Tailwind CSS)"]
        WS["WebSocket Client<br/>(LiveView)"]
        XTerm["xterm.js<br/>(Terminal)"]
    end
    subgraph Backend["Django Backend"]
        API["REST API<br/>(DRF + Ninja)"]
        LV["LiveView Server<br/>(Channels)"]
        Celery["Celery Workers"]
        Auth["Auth & RBAC"]
        DB[("PostgreSQL")]
        Cache[("Redis")]
    end
    subgraph Agent["OpenClaw Agent Container"]
        GW["Gateway Server"]
        Plugin["openclaw-kanban<br/>Plugin"]
        Shell["Shell Server<br/>(PTY Bridge)"]
        AI["Claude / Mistral<br/>LLM"]
    end
    subgraph Infra["Infrastructure"]
        OVH["OVH Cloud<br/>(K8s / Bare-metal)"]
    end
    UI -->|HTTPS| API
    WS -->|WebSocket| LV
    XTerm -->|WebSocket| Shell
    API -->|SQL| DB
    API -->|Cache| Cache
    Celery -->|SQL| DB
    Celery -->|Redis Queue| Cache
    Celery -->|Dispatch Jobs| Plugin
    Plugin -->|REST API| API
    Plugin -->|LLM Calls| AI
    Backend -->|Runs on| OVH
    

Data Model Overview

classDiagram
    direction BT
    class User {
        +int id
        +str username
        +str email
        +bool is_superuser
    }
    class UserProfile {
        +bool is_bot
        +str agent_identifier
        +str avatar
    }
    class Board {
        +int id
        +str name
        +str slug
        +bool sprints_enabled
        +bool is_leads_board
        +User owner
    }
    class Column {
        +int id
        +str name
        +int position
        +Board board
    }
    class Task {
        +int id
        +str title
        +text description
        +str status
        +str priority
        +date due_date
        +Column column
        +Epic epic
        +Sprint sprint
    }
    class Epic {
        +int id
        +str name
        +str color
        +Board board
    }
    class Sprint {
        +int id
        +str name
        +date start_date
        +date end_date
        +Board board
    }
    class ChatMessage {
        +int id
        +str role
        +text content
        +Board board
        +User sender
    }
    class AgentDispatchJob {
        +int id
        +str agent_identifier
        +str status
        +json context
        +json result
    }
    User --> UserProfile : "has"
    User --> Board : "owns"
    Board --> Column : "has many"
    Board --> Epic : "has many"
    Board --> Sprint : "has many"
    Column --> Task : "contains"
    Epic --> Task : "groups"
    Sprint --> Task : "schedules"
    Board --> ChatMessage : "stores"
    Task --> Task : "subtasks"
    User --> Task : "assigned to"
    

Key Features

Business Use Cases

• AI‑Assisted Project Planning – Generate task breakdowns, timelines and risk analysis from plain‑language briefs.
• Workflow Automation – Trigger‑based automations for recurring processes with AI‑suggested optimisations.
• Secure Collaboration – Real‑time board updates, integrated chat, and future audio/video tools.
• GDPR‑First Architecture – EU‑hosted data storage, right‑to‑erasure tooling and full audit logs.
• European Business Integrations – Connectors for Pennylane, Chorus Pro, Dolibarr and other EU ERP systems.
• AI Agent Orchestration – Deploy configurable AI agents to draft, research, schedule or execute tasks.

Screenshots

Home dashboard showing all boards and recent activity

Board in columns (Kanban) view with tasks

Task detail modal with fields, subtasks and comments