█▀█ █▀▀ █▀▀ █▄░█ ▀█▀   █▄▄ █▀█ █▀█ █▀█ █▀▄
                            █▀█ █▄█ ██▄ █░▀█ ░█░   █▄█ █▄█ █▀█ █▀▄ █▄▀

Opensource TUI based Kanban board for orchestrating AI coding agents

AgentBoard is a terminal-based Kanban board for orchestrating and managing AI coding agents. It provides a visual development workflow for modern software engineering, bridging project management with automated code generation.

---

Key Features

• Modern Kanban TUI: Terminal UI built with Bubble Tea, featuring glassmorphism-inspired borders and smooth animations.
• Agent Orchestration: Spawn and manage agents like Claude Code, OpenCode, and Cursor via tmux panes or embedded PTY.
• MCP Native: Integrated Model Context Protocol support via @thisisayande/contextcarry-mcp for agent memory persistence.
• Persistent Storage: SQLite backend with automatic migrations.
• AI Proposal Flow: Coordinator LLM creates worker prompts, with approval gating before execution.
• LangChain Go Integration: Provider registry supporting OpenAI, Ollama, Claude, and Zai.

---

🌐 Marketing Website

For more information about AgentBoard, including feature showcases, screenshots, and project updates, visit the official website:

https://agentboard.ayande.xyz/

---

Getting Started

Prerequisites

• Go: 1.21 or higher
• git: For cloning the repository
• tmux: Required for tmux-mode agent spawning (Linux/macOS)
• Node.js/npm: Required for MCP server integrations

Installation

From Binary (All Platforms)

Run the automated installer:

curl -sSL https://raw.githubusercontent.com/ayan-de/agent-board/main/install.sh | bash

This will detect your OS and architecture, install dependencies, and place the binary in ~/.local/bin.

From Source

Ensure Go 1.21+ is installed.

git clone https://github.com/ayan-de/agent-board.git
cd agent-board
go build -o agentboard ./cmd/agentboard

# Add to PATH (optional)
export PATH="$HOME/.local/bin:$PATH"

# Run it in the project you want to use it in
./agentboard

Platform-Specific Notes

Platform	Package Manager	Install Command
Linux (apt)	apt-get install golang git tmux npm	Binary or source
macOS (brew)	brew install go git tmux npm node	Binary or source
Android (Termux)	pkg install golang git tmux nodejs	Binary or source

Basic Usage

Start the interactive Kanban board:

./agentboard

Or if installed via binary:

agentboard

CLI Commands

Command	Description
agentboard	Launch the Kanban TUI (default)
agentboard update	Check for and install the latest release in-place
agentboard -v, --version, version	Print the installed version
agentboard -h, --help, help	Show usage and available commands

Keybindings

Key	Action
h/l or ←/→	Move between Kanban columns
j/k or ↑/↓	Move between tickets
Enter	Open ticket detail view
a	Add a new ticket in the active column
d	Delete selected ticket
1-4	Jump to a specific column
?	Toggle help view
i	Toggle agent dashboard
:	Open command palette
q	Quit with confirmation
Esc	Return to board from other views

Ticket View

Key	Action
j/k or ↑/↓	Move between fields
e	Edit the selected editable field
s	Cycle ticket status
a	Open agent selection
p	Approve pending proposal
r	Start approved run
Esc	Cancel edit or return to board

Dashboard

Key	Action
r	Re-run agent detection
Esc	Return to board

---

Configuration

Configuration is loaded from ~/.agentboard/config.toml (global) or <project>/.agentboard/config.toml (project-level).

Example Configuration

[general]
log = "info"
tmux = true

[db]
path = ".agentboard/agentboard.db"

[llm]
provider = "openai"
model = "gpt-4"
api_key = "${OPENAI_API_KEY}"

[llm.coordinator]
model = "gpt-4"

[llm.summarizer]
model = "gpt-4"

[tui]
theme = "nord"

[mcp.contextcarry]
command = "npx"
args = ["-y", "@thisisayande/contextcarry-mcp"]

Environment Variables

Override config values:

Variable	Config Key
AGENTBOARD_LOG	general.log
AGENTBOARD_TMUX	general.tmux
AGENTBOARD_DB	db.path
AGENTBOARD_LLM_PROVIDER	llm.provider
AGENTBOARD_LLM_MODEL	llm.model
AGENTBOARD_LLM_API_KEY	llm.api_key
AGENTBOARD_LLM_BASE_URL	llm.base_url
AGENTBOARD_LLM_COORDINATOR_MODEL	llm.coordinator_model
AGENTBOARD_LLM_SUMMARIZER_MODEL	llm.summarizer_model
AGENTBOARD_NPM_PATH	mcp.npm_path
AGENTBOARD_NODE_PATH	mcp.node_path

---

Themes

AgentBoard ships with builtin themes. User themes can be placed in ~/.agentboard/themes/ or <project>/.agentboard/themes/ as JSON files.

Available Themes

Theme	Description
agentboard	Default AgentBoard theme
catppuccin	Soft, warm pastel
dracula	Dark with vibrant colors
gruvbox	Retro groove terminal
matrix	Cyberpunk green-on-black
nord	Clean, arctic palette
tokyonight	Tokyo night city lights

---

Architecture

agent-board/
├── cmd/agentboard/
│   └── main.go             # TUI entrypoint
├── internal/
│   ├── tui/                # Bubble Tea application
│   ├── store/              # SQLite persistence
│   ├── config/             # Config loading and agent detection
│   ├── theme/              # Theme registry
│   ├── keybinding/         # Keymap and action resolution
│   ├── llm/                # LangChain Go provider registry
│   ├── orchestrator/       # Agent lifecycle management
│   ├── pty/                # PTY agent configurations
│   ├── prompt/             # LLM prompt templates
│   ├── mcp/                # MCP manager and adapters
│   └── mcpclient/          # MCP stdio client wrapper
├── docs/                   # Design notes
└── AGENTS.md               # Project state documentation

---

Development

Prerequisites

• Go: 1.21+
• tmux: Required for agent spawning
• Node.js/npm: Required for MCP server integrations

Commands

# Build
go build -o agentboard ./cmd/agentboard

# Run tests
go test ./...

# Run vet
go vet ./...

Debug Mode

AGENTBOARD_LOG=debug ./agentboard

3. Running Tests: We follow a strict TDD discipline. Ensure all tests pass before submitting changes.

   go test ./...

4. Database Migrations: Migrations are handled automatically on startup, but you can inspect the schema in internal/store/migrations.go.

---

🤝 Contributing

We welcome contributions! Please follow our workflow to keep the board clean and efficient:

1. Fork the repository and create your feature branch.
2. Write Tests First: We strictly follow Red-Green-Refactor. No implementation without a corresponding _test.go file.
3. Lint your code: Run go vet ./... to ensure idiomatic Go.
4. Document: Update AGENT.md if you introduce architectural changes.
5. Submit PR: Ensure your commit messages follow Conventional Commits.

Architecture Overview

• cmd/: Application entrypoints.
• internal/tui/: Bubble Tea models and components.
• internal/orchestrator/: Agent lifecycle and process management.
• internal/store/: SQLite persistence layer.
• internal/mcp/: Protocol clients for context management.

---

📜 License

Distributed under the MIT License. See LICENSE for more information.

---

Built with ❤️ for the AI-First Engineering community.

---

Contributing

1. Fork and create a feature branch from main
2. Write tests first (TDD discipline)
3. Make tests pass
4. Refactor while keeping tests green
5. Run go test ./... and go vet ./...
6. Update AGENTS.md for architectural changes
7. Submit a PR with Conventional Commits

---

License

MIT License. See LICENSE for details.