Skip to content

dashboard.md

Latest commit

 

History

History
240 lines (155 loc) · 8.85 KB

dashboard.md

File metadata and controls

240 lines (155 loc) · 8.85 KB
Raw
title Admin Dashboard
description Manage the JD.AI Gateway through the Blazor WebAssembly dashboard with real-time monitoring, agent management, and configuration UI.

Admin Dashboard

The JD.AI Admin Dashboard is a Blazor WebAssembly application that provides a real-time management interface for the JD.AI Gateway. It connects to the Gateway via REST API and SignalR for live updates.

Dashboard overview page

Getting Started

  1. Start the Gateway:

    cd src/JD.AI.Gateway
dotnet run

  2. Open the dashboard at http://localhost:5189/

The top-right corner shows a LIVE indicator when the dashboard is connected to the Gateway via SignalR.

Overview Page

The Overview page provides a high-level summary of the Gateway state:

  • Agents — number of currently running agent instances
  • Channels — active communication channels (Discord, Signal, WebChat, etc.)
  • Sessions — active chat sessions
  • OpenClaw — bridge connection status

The Recent Activity feed displays real-time events as they occur, including agent spawns, channel connections, and message activity.

Real-Time Session Monitoring

The dashboard streams live events via SignalR:

  • Agent turn completions with token counts
  • Channel message arrivals and deliveries
  • Provider status changes
  • Update notifications

Connection status is shown in the top-right corner with a green "LIVE" indicator. If the connection drops, the dashboard automatically reconnects.

Agent Management

The Agents page lists all currently running agent instances with full lifecycle controls.

Agents management page

Column Description
ID Unique agent instance identifier
Provider AI provider (ollama, claude-code, github-copilot)
Model Model being used (e.g., qwen3.5:27b)
Turns Number of conversation turns completed
Created When the agent was spawned

Spawning agents

Click + Spawn Agent to create a new agent instance. Configure:

  • Provider and model selection
  • System prompt
  • Model parameters (temperature, top-p, max tokens)

Stopping agents

Use the trash icon to stop and remove an agent. Active sessions are gracefully drained before removal.

Agent monitoring

From the Agents page, monitor each agent's:

  • Token consumption — cumulative prompt and completion tokens
  • Turn latency — average response time per turn
  • Provider health — whether the backing provider is reachable
  • Active session — the current session associated with the agent

Channel Status

The Channels page shows all configured communication channels and their connection status.

Channel status and connectivity controls

Each channel card displays:

  • Channel name and type (web, discord, signal, slack, telegram)
  • Connection status (Online / Offline)
  • Disconnect / Connect controls
  • Override — configure channel-specific settings

The Sync OpenClaw button synchronizes channel configuration with the OpenClaw bridge.

Web Chat

The Chat page provides a built-in web interface for communicating with agents:

Web chat with streaming responses

  • Agent selector — choose which agent to chat with from the dropdown
  • Streaming responses — tokens stream in real-time via SignalR
  • Message history — full conversation displayed with user/assistant bubbles
  • Markdown rendering — code blocks and formatting rendered inline
  • Clear conversation — reset the chat at any time

Messages are sent to the Gateway's AgentHub and streamed back as chunked responses.

Configuration UI

The Settings page provides a WYSIWYG editor for all Gateway configuration. Changes are applied in real-time and persisted to the configuration file.

Settings editor

Tip

Changes made through the Settings UI are validated before applying. Invalid values are rejected with a descriptive error message.

Server settings

  • Network — host and port binding
  • Behavior — verbose logging toggle
  • Authentication — enable/disable API key authentication
  • Rate Limiting — toggle rate limiting and set max requests per minute

Provider settings

  • Add/remove providers
  • Configure endpoints and API keys
  • Set default models per provider

Agent settings

  • Agent name, provider, and model selection
  • System prompt customization
  • Model parameters: temperature, top-p, top-k, max tokens, context window, penalties, seed, stop sequences

Channel settings

Configure channel adapters (Discord, Signal, Slack, Telegram, WebChat) with per-channel credentials and options.

Each channel configuration includes:

  • Bot token / credentials — channel-specific authentication
  • Target channels / groups — where the bot listens and responds
  • Message format — how agent responses are formatted for the platform
  • Auto-connect — whether the channel connects automatically on gateway startup

Routing settings

  • Default agent assignment
  • Channel-to-agent mappings
  • Routing strategy selection

Model Providers

The Providers page shows all configured AI providers, their connection status, and available models. Each provider card shows:

Provider catalog and status page

  • Provider name and status (Online / Offline)
  • Number of available models
  • Model catalog with ID and display name

Supported providers include Claude Code, GitHub Copilot, OpenAI Codex, Ollama, local GGUF models (via LLamaSharp), and API-key providers (OpenAI, Anthropic, Google Gemini, Mistral, AWS Bedrock, HuggingFace, and OpenAI-compatible endpoints).

Routing Configuration

The Routing tab in Settings defines how messages are routed to agents:

Routing configuration page

  • Default agent — which agent handles messages when no specific route matches
  • Channel-to-agent mappings — assign specific agents to specific channels
  • Routing strategy — round-robin, priority-based, or direct assignment

OpenClaw Bridge

The OpenClaw tab configures cross-gateway orchestration:

  • Bridge URL and connection settings
  • Agent registration for remote access
  • Channel override mappings for multi-gateway deployments

Authentication

When Gateway:Auth:Enabled is true, the dashboard requires authentication. Pass the API key via the login prompt on first access.

Dashboard access follows the same role-based model as the REST API:

Role Dashboard capabilities
User View-only access to all pages
Operator Send messages, connect/disconnect channels
Admin Full access including agent lifecycle and settings

See Security for details on configuring API keys and roles.

Architecture

┌─────────────────┐     REST API      ┌──────────────────┐
│   Dashboard      │ ◄──────────────► │   Gateway API    │
│   (Blazor WASM)  │     SignalR       │   (ASP.NET Core) │
│                  │ ◄──────────────► │                  │
│  - Pages/        │                   │  - Endpoints/    │
│  - Components/   │                   │  - Hubs/         │
│  - Services/     │                   │  - Services/     │
└─────────────────┘                   └──────────────────┘

The dashboard is served as static files from the Gateway and communicates entirely through HTTP and WebSocket connections.

Deployment considerations

  • The dashboard is a static Blazor WASM app bundled with the Gateway — no separate deployment required
  • For production, place the Gateway behind a TLS-terminating reverse proxy (see Deployment)
  • Enable Auth.Enabled to require authentication for both API and dashboard access
  • The dashboard works on any modern browser (Chrome, Firefox, Edge, Safari)

Troubleshooting

Dashboard shows "Disconnected"

The SignalR connection to the gateway was lost. Check that:

  • The gateway process is running: jdai-daemon status
  • The gateway port (default 18789) is accessible
  • No firewall rules block WebSocket connections

Settings changes not persisting

Ensure the gateway process has write access to appsettings.json. On Linux, check file permissions.

See also