Skip to content

baconfy/smart-management

Folders and files

NameName
Last commit message
Last commit date

Latest commit

Β 

History

128 Commits
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 

Repository files navigation

Smart Management

Talk to AI. Project management happens.

An open-source, AI-powered project manager where conversations replace forms, kanbans, and dashboards. You describe your project. AI agents organize everything behind the scenes.

Quick Start β€’ Features β€’ How It Works β€’ Tech Stack β€’ Contributing

License Laravel 12 React 19 Tests Multi-provider AI


The Problem

Project management tools treat management as bureaucracy β€” you feed the system so it can work. It's work to generate work.

Using an AI chat as a project manager works surprisingly well β€” it's conversational, natural, zero friction. Until the chat hits its limits and dies. The conversation is lost, and with it, all the context, decisions, and structure that had been built.

Smart Management fixes this. The conversational approach works. What was missing is persistence, structure, and scoped context.


✨ Features

πŸ’¬ Chat-First Interface

The primary interface is a conversation, not a dashboard. Describe your project, discuss decisions, define rules β€” AI agents organize everything behind the scenes. Traditional views (kanban, decisions, business rules) exist as read-only visualizations of what the AI has already organized.

πŸ€– Multi-Agent System

A team of specialized AI agents, each with a distinct role:

Agent Purpose Creates
Architect Stack choices, patterns, tradeoffs Decisions (ADRs)
Project Manager Tasks, roadmap, estimates, progress Tasks & Roadmap
Business Analyst Domain rules, requirements, acceptance criteria Business Rules
Technical Implementation help, code review, debugging Implementation Notes

You never pick which agent responds β€” an invisible Moderator (running on a cheap model) classifies your message and routes it to the right specialist. When it's unsure, it asks you.

πŸ“š Knowledge That Outlives Conversations

Agents don't share chat history. They read and write structured artifacts: Decisions, Business Rules, Tasks, and Implementation Notes. Conversations can be deleted without losing project knowledge. Your artifacts are your project's brain.

πŸ“‹ Kanban Board

Drag-and-drop task board with customizable statuses per project. Tasks are created by the PM agent during conversations, but you can also view and reorganize them visually.

βš™οΈ Per-Project Agent Customization

Each project gets its own set of agents. Edit instructions to fine-tune behavior:

  • Make the Architect emphasize security for a crypto project
  • Make the PM more aggressive on deadlines for a tight sprint
  • Reset to default anytime β€” you can't permanently break an agent

πŸ”‘ Bring Your Own API Keys

Multi-provider support via Laravel AI SDK. Use Anthropic, OpenAI, Gemini, Groq, DeepSeek, Mistral, xAI, Ollama (local), and more. You own your keys, your data, your infrastructure.

πŸ§ͺ 427+ Tests

Comprehensive test suite covering the entire backend: models, actions, services, controllers, AI tools, and real-time events.


βš™οΈ How It Works

πŸ“ Project
β”œβ”€β”€ πŸ’¬ Conversations (the meeting room)
β”‚   β”œβ”€β”€ "Defining stack and approach"
β”‚   β”œβ”€β”€ "Payment split rules"
β”‚   └── "Replanning phase 2"
β”œβ”€β”€ πŸ“š Artifacts (generated from conversations)
β”‚   β”œβ”€β”€ Decisions: "Use AdonisJS", "Non-custodial"
β”‚   β”œβ”€β”€ Business Rules: "Immediate split", "0.5% fee"
β”‚   └── Tasks & Roadmap
└── πŸ“‹ Kanban Board
    └── Drag-and-drop task management

The Meeting Room model: You open a conversation and start talking. The Moderator classifies your message and routes it to the right agent (or multiple agents). Each agent reads relevant artifacts for context and writes new ones as output.

You:        "Should the payment split happen on-chain or off-chain?"

Moderator:  β†’ Business + Architecture question β†’ routes to Analyst + Architect

You see:    A tabbed response from both agents,
            each covering their perspective.

Task Technical Chat: Click a task, start a conversation, and a Technical agent joins with full context about the task, related decisions, and business rules β€” ready to help you implement.


πŸ›  Tech Stack

Layer Technology
Backend Laravel 12, PHP 8.5
Frontend React 19, Inertia.js v2, Tailwind CSS v4
AI Laravel AI SDK (multi-provider)
Real-time Laravel Reverb (WebSockets)
Queue Laravel Horizon + Redis
Database PostgreSQL 18
Storage MinIO (S3-compatible)
Auth Laravel Fortify (with 2FA)
Testing Pest 4
Infrastructure Docker (serversideup/php)

πŸš€ Quick Start

Prerequisites

1. Clone the repository

git clone https://github.com/your-username/smart-management.git
cd smart-management

2. Copy the environment file

cp .env.example .env

3. Configure your AI provider(s)

Open .env and add at least one API key:

# Add one or more AI provider keys
ANTHROPIC_API_KEY=sk-ant-...
OPENAI_API_KEY=sk-...
GEMINI_API_KEY=...

# All supported providers:
# ANTHROPIC_API_KEY, OPENAI_API_KEY, GEMINI_API_KEY,
# GROQ_API_KEY, DEEPSEEK_API_KEY, MISTRAL_API_KEY,
# XAI_API_KEY, OLLAMA_API_KEY, OPENROUTER_API_KEY,
# COHERE_API_KEY, VOYAGEAI_API_KEY, JINA_API_KEY

4. Start the development environment

composer dev

This single command boots everything: PHP, PostgreSQL, Redis, Horizon, Reverb, MinIO, Mailpit, and Vite with hot-reload.

5. Install dependencies and setup

In another terminal:

# Install PHP dependencies
docker compose exec app composer install

# Generate application key
docker compose exec app php artisan key:generate

# Run migrations
docker compose exec app php artisan migrate

# Install frontend dependencies and build
docker compose exec app npm install
docker compose exec app npm run build

6. Open the app

Visit http://localhost (or the port configured in FORWARD_APP_PORT).

Register an account, create your first project, and start chatting!


πŸ“ Project Structure

app/
β”œβ”€β”€ Actions/            # Single-responsibility invokable classes
β”œβ”€β”€ Ai/
β”‚   β”œβ”€β”€ Agents/         # GenericAgent (unified for all types)
β”‚   β”œβ”€β”€ Stores/         # ProjectConversationStore
β”‚   └── Tools/          # 16 AI tools (CRUD for each artifact type)
β”œβ”€β”€ Enums/              # AgentType, DecisionStatus, TaskPriority, etc.
β”œβ”€β”€ Events/             # AgentMessageReceived, AgentsProcessing, etc.
β”œβ”€β”€ Models/             # Eloquent models
└── Services/           # Orchestrators (CreateProjectService, etc.)

resources/
β”œβ”€β”€ instructions/       # Default agent instructions (.md files)
└── js/
    └── pages/
        └── projects/
            β”œβ”€β”€ conversations/   # Chat interface
            β”œβ”€β”€ tasks/           # Kanban board + task detail
            β”œβ”€β”€ decisions/       # Post-it grid view
            β”œβ”€β”€ business-rules/  # Accordion view
            └── settings/        # Agent management

🐳 Docker Services

Service Description
app Laravel + Nginx
horizon Queue worker
reverb WebSocket server
postgres PostgreSQL 18
redis Cache + Queue + Sessions
minio S3-compatible storage
mailpit Email testing

πŸ”§ Configuration

AI Provider

The default AI provider and model are configured in config/ai.php. You can override the model per agent in the project settings UI.

Cost tip: The invisible Moderator uses the cheapest available model automatically (e.g., Claude Haiku, GPT-4o-mini) since it only classifies messages. Your budget goes to the agents that matter.

WebSockets (Reverb)

The default configuration works out of the box with Docker. If deploying to production, update the Reverb variables in .env:

REVERB_HOST=your-domain.com
REVERB_PORT=443
REVERB_SCHEME=https

VITE_REVERB_HOST="${REVERB_HOST}"
VITE_REVERB_PORT="${REVERB_PORT}"
VITE_REVERB_SCHEME="${REVERB_SCHEME}"

Running Tests

docker compose exec app php artisan test --compact

πŸ—Ί Roadmap

See ROADMAP.md for the full phased execution plan.

Completed:

  • βœ… Phase 1 β€” Foundation (projects, agents, chat, decisions)
  • βœ… Phase 1.5 β€” Real-time chat (WebSockets, parallel agents)
  • βœ… Phase 2 β€” Full agent system (all 4 agents + all artifact types)
  • βœ… Phase 3 β€” Task system (kanban, technical chat, unified chat)
  • βœ… Phase 4.1 β€” Settings & agent management

Coming next:

  • πŸ”„ Token streaming (SSE)
  • πŸ”„ Custom agent creation with AI-assisted instructions
  • πŸ”„ Project dashboard & overview
  • πŸ”„ Context window management
  • πŸ”„ Integrations (GitHub, Slack)

🀝 Contributing

Contributions are welcome! This is a community-driven project.

  1. Fork the repository
  2. Create a feature branch (git checkout -b feature/amazing-feature)
  3. Write tests for your changes (we maintain 427+ tests β€” keep it that way)
  4. Run the test suite (php artisan test --compact)
  5. Run the code formatter (vendor/bin/pint --dirty)
  6. Commit and push
  7. Open a Pull Request

Please read VISION.md to understand the architectural decisions before contributing.


β˜• Support the Project

If Smart Management is useful to you, consider buying me a coffee! It helps keep the project alive and motivated.

Buy Me A Coffee


πŸ“„ License

Smart Management is open-source software released under a non-commercial license.

You are free to:

  • βœ… Use it for personal projects
  • βœ… Self-host for your own use
  • βœ… Study, modify, and learn from the code
  • βœ… Contribute to the project

You may not:

  • ❌ Use it for commercial purposes
  • ❌ Sell it or offer it as a paid service (SaaS)
  • ❌ Redistribute modified versions for commercial gain

⭐ Star History

If Smart Management helped you manage a project (or just sparked an idea), consider giving it a ⭐ on GitHub. It helps others discover the project.



Built with ❀️ by the community.

"The user talks to the AI, and project management happens as a consequence."

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages