AiTasker

A lightweight task and project management server built in Go, designed for AI agent integration via the Model Context Protocol (MCP).

Overview

AiTasker exposes project and task management capabilities as MCP tools, allowing AI agents (like Claude) to programmatically create projects, manage tasks, and track work through a standard HTTP/JSON-RPC interface.

Features

• Project Management — Create, list, retrieve, and delete projects
• Task Management — Full CRUD for tasks within projects, with status tracking
• Task Workflow — Four-state lifecycle: TODO → IN_PROGRESS → IN_REVIEW → DONE
• Comments — Add and retrieve comments on tasks
• MCP Integration — 11 tools exposed via JSON-RPC 2.0 for AI agent use
• File-based Storage — No database required; data persisted as JSON files
• Thread-safe — Concurrent-safe repository operations via sync.RWMutex

Tech Stack

• Language: Go 1.25+
• HTTP Router: gorilla/mux
• ID Generation: google/uuid
• Protocol: Model Context Protocol v2024-11-05

Project Structure

AiTasker/
├── cmd/server/main.go           # Entry point
├── internal/
│   ├── delivery/
│   │   ├── http/router.go       # HTTP routing
│   │   └── mcp/mcp_server.go   # MCP protocol handler
│   ├── features/
│   │   ├── project/             # Project domain & service
│   │   ├── task/                # Task domain & service
│   │   └── comment/             # Comment domain
│   └── shared/                  # Errors, time provider
├── go.mod
└── go.sum

Getting Started

Prerequisites

• Go 1.25 or later

Installation

git clone https://github.com/your-username/AiTasker.git
cd AiTasker
go mod download

Running

go run ./cmd/server

Or build and run:

go build -o aitasker ./cmd/server
./aitasker

The server starts on http://localhost:8080.

Testing

go test ./...

MCP Tools

The server exposes the following tools via POST /mcp:

MCP Request Format

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "create_project",
    "arguments": {
      "name": "MyProject",
      "description": "Project description"
    }
  }
}

Data Storage

Projects and tasks are stored as JSON files under ./data/:

data/
├── MyProject/
│   ├── meta.json      # Project metadata
│   ├── MY-1.json      # Task files (prefix = first 2 chars of project name)
│   └── MY-2.json

Connecting to Claude

Add AiTasker as an MCP server in your Claude configuration:

{
  "mcpServers": {
    "aitasker": {
      "url": "http://localhost:8080/mcp"
    }
  }
}

Claude will then have access to all 11 task management tools directly within conversations.

License

MIT