Skip to content

Mowd/claude_dashboard

Repository files navigation

English | 繁體中文

Claude Dashboard

Multi-agent software development dashboard powered by Claude.

Claude Dashboard Screenshot

Features

  • Multi-agent pipeline — Automated PM → RD → UI → TEST → SEC workflow, each agent with dedicated responsibilities and tools
  • Real-time streaming output — Token-by-token display of agent reasoning and actions via WebSocket
  • Interactive terminal — Full-featured terminal (xterm.js + node-pty) for live command inspection
  • Workflow management — Pause, resume, and cancel running workflows at any point
  • Event log & token tracking — Monitor agent activities, tool usage, and token consumption per step
  • SQLite persistence — All workflows, agent outputs, and metrics are persisted for later review

Tech Stack

Layer Technology
Frontend Next.js 15, React 19, Tailwind CSS, Zustand
Backend Express, WebSocket (ws), node-pty
Database sql.js (SQLite WASM)
AI Claude CLI

Prerequisites

  • Platform — Windows, macOS, Linux
  • Node.js 23.6+
  • Claude CLI installed and configured (docs)

Quick Start

Install via npm

npm install -g claude-dashboard
cd /path/to/your/project
cdb

Install from source

git clone https://github.com/Mowd/claude_dashboard.git
cd claude_dashboard
npm install
npm link
cd /path/to/your/project
cdb

Agent Roles

Role Responsibility Tools Timeout
PM Analyze requirements, produce structured specifications Read 3 min
RD Design backend architecture, implement server-side code Read, Edit, Bash 10 min
UI Design and implement frontend components, pages, styling Read, Edit, Bash 10 min
TEST Write and execute tests, report coverage and bugs Read, Edit, Bash 10 min
SEC Security assessment, vulnerability scanning, audit Read, Bash 5 min

Agents execute sequentially. Each agent receives the accumulated context from all prior agents.

Project Structure

claude_dashboard/
├── bin/                  # CLI launcher (cdb)
├── prompts/              # Agent system prompt templates
│   ├── pm-system.md
│   ├── rd-system.md
│   ├── ui-system.md
│   ├── test-system.md
│   └── sec-system.md
├── server.ts             # Express + WebSocket server entry
├── src/
│   ├── app/              # Next.js pages & API routes
│   ├── components/       # React components
│   │   ├── agent/        #   Agent cards & output display
│   │   ├── events/       #   Event log
│   │   ├── history/      #   Workflow history table
│   │   ├── layout/       #   Shell & navigation
│   │   ├── pipeline/     #   Pipeline visualization
│   │   ├── terminal/     #   xterm.js terminal panel
│   │   ├── ui/           #   Shared UI primitives
│   │   └── workflow/     #   Workflow launcher
│   ├── hooks/            # Custom React hooks
│   ├── lib/
│   │   ├── agents/       #   Prompt loading & construction
│   │   ├── db/           #   SQLite schema, connection, queries
│   │   ├── terminal/     #   PTY manager
│   │   ├── websocket/    #   WebSocket server & protocol
│   │   └── workflow/     #   Engine, pipeline, agent runner
│   ├── stores/           # Zustand state stores
│   └── types/            # TypeScript type definitions
├── data/                 # SQLite database files
├── package.json
└── tsconfig.json

Changelog (0.1.0 → 0.5.0)

0.5.0

  • Added tool use summaries in agent streaming output — live display of which tools (Read, Edit, Bash, etc.) each agent is invoking with key parameters.
  • Fixed horizontal scrollbar overflow across all pages by adding overflow constraints at multiple CSS and component levels (prose word-break, grid clipping, vertical-only scroll on history pages).
  • Improved agent completion state — clean final output replaces streaming chunks so tool summaries are shown during execution but removed from the finished result.

0.4.0

  • Added selectable execution plans (Full / Fast / Custom) so small tasks can skip TEST/SEC when appropriate.
  • Added Impact Preview before run start to estimate risk and suggest a run mode.
  • Delivered broad i18n support (English + Traditional Chinese) with in-app language switcher and localized UI copy across dashboard, history, terminal, status labels, and events.

0.3.0

  • Introduced Workflow History and Workflow Detail pages backed by SQLite persistence.
  • Added history filters, pagination, retention cleanup, and workflow metrics.
  • Added templates, artifact summary, and retry-as-new-run UX.
  • Hardened API/DB initialization paths and fixed history 500 issues.
  • Improved terminal reliability with route-switch reconnect/reattach fixes, including TUI (htop) restoration and replay buffering.

0.2.0

  • Stabilized npm package publishing and global install flow (cdb launcher path and root detection fixes).
  • Improved cross-platform compatibility by migrating DB runtime to sql.js (SQLite WASM).
  • Migrated server/tooling to native ESM and standardized test suite on bun:test.
  • Updated installation and quick-start documentation.

0.1.0

  • Initial public baseline of Claude Dashboard:
    • multi-agent PM→RD→UI→TEST→SEC pipeline
    • real-time WebSocket streaming output
    • integrated interactive terminal (xterm.js + node-pty)
    • core workflow controls (start/pause/resume/cancel)

☕ Buy Me a Coffee

If you find this project helpful, consider buying me a coffee!

Buy Me A Coffee

License

MIT

About

Multi-agent software development dashboard powered by Claude Code.

Resources

License

Stars

Watchers

Forks

Packages

 
 
 

Contributors