Tareas

A self-hosted task and project management tool built with FastAPI and Vanilla JavaScript. Tareas is optimized for collaboration with AI agents: agents can work with projects, subtasks, notes, dependencies, and assignments through the built-in MCP server while all write actions stay attributable and auditable.

Features

• Task & Project Management - Create tasks, organize them into projects with subtasks, dependencies, and deadlines
• Interactive Network Diagram - Visualize project dependencies as an interactive graph (vis-network)
• Team Collaboration - Assign tasks, manage team permissions (read/edit/create), notes system
• WYSIWYG Editor - Rich text descriptions for tasks and subtasks
• File Storage - Nextcloud/WebDAV integration with tree view, drag & drop upload, context menus
• ONLYOFFICE Integration - Edit Office documents (docx, xlsx, pptx) directly in the browser via WOPI
• LDAP/Active Directory - Authenticate users against AD, automatic sync, group-based access
• Email Notifications - SMTP integration with configurable templates for assignments, status changes, deadlines
• AI Agent Collaboration - Built-in MCP server for project/task automation by AI agents with token-based access
• TLS Support - Optional HTTPS with certificate management via Admin UI
• Admin Panel - Separate admin interface (Port 8505) for user management, LDAP, Nextcloud, ONLYOFFICE, mail, TLS, and MCP configuration
• Light/Dark Theme - CSS Custom Properties with persistent preference
• APT Package - Install via .deb package on Ubuntu/Debian

Screenshots

Project view — Rich text description, Nextcloud file browser, inline editing

Subtask list — Dependencies, status tracking, priority, assignments

Network diagram — Interactive dependency graph with auto-layout and status colors

Installation

Option 1: APT Package (Recommended)

curl -s https://your-repo-server/tareas/install.sh | bash

This sets up the APT repository and installs Tareas with all dependencies.

Option 2: Manual Installation

# Clone the repository
git clone https://github.com/Manerba/tareas.git
cd tareas

# Create virtual environment
python3 -m venv venv
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Start the application
python dashboard/app.py

The dashboard is available at http://localhost:8504.

Note: Tested on Ubuntu 24.04 Server (standard installation).

Option 3: Systemd Service

After manual installation, copy the service files:

cp scripts/tareas.service /etc/systemd/system/
cp scripts/tareas-admin.service /etc/systemd/system/
systemctl daemon-reload
systemctl enable --now tareas tareas-admin

Configuration

All configuration is managed through the Admin Panel at http://localhost:8505:

Feature	Admin Tab	Description
General	Allgemein	Server address for email links
Users	Benutzer	Create/manage local users
LDAP	LDAP	Active Directory connection and sync
Nextcloud	Nextcloud	WebDAV file storage integration
ONLYOFFICE	ONLYOFFICE	Document Server URL and JWT secret
Mail	Mail	SMTP settings and notification templates
TLS	TLS	HTTPS certificate paths
MCP	MCP	Server status, API tokens, audit log, global kill switch

MCP Server

Tareas exposes a Model Context Protocol (MCP) server at /mcp/ for AI agents and remote coding assistants. The server is mounted in the main app on port 8504 and uses Streamable HTTP via FastMCP.

• Authentication uses Authorization: Bearer <token>.
• MCP tokens are created in the Admin Panel under MCP and are shown only once.
• Tokens are stored as SHA-256 hashes; revoked tokens stop working immediately.
• Every token is mapped to its own mcp user in the database.
• Write operations are recorded in the audit log and can be reviewed in the Admin Panel.
• Available tools cover projects, subtasks, dependencies, notes, users, areas, search, self-assignment, and whoami.

Default Credentials

On first start, create an admin user via the admin panel. The first user created automatically gets admin privileges.

Database

Tareas uses SQLite with WAL mode. The database is stored at data/tareas.db and created automatically on first start. No external database server required.

Architecture

Tareas/
├── dashboard/
│   ├── app.py              # Main FastAPI app (Port 8504)
│   ├── admin_app.py         # Admin FastAPI app (Port 8505)
│   ├── api_*.py             # API routers (tasks, auth, teams, etc.)
│   ├── mcp_server.py        # MCP tools for AI agents
│   ├── components/          # Reusable Python components
│   ├── static/
│   │   ├── css/             # Stylesheets (theme, components)
│   │   └── js/              # Frontend JavaScript (SPA)
│   └── templates/           # HTML templates
├── scripts/                 # Systemd service files
├── packaging/               # .deb package configuration
├── data/                    # SQLite database (created at runtime)
├── requirements.txt
└── version.txt

Tech Stack

Layer	Technology
Backend	Python 3.10+, FastAPI, Uvicorn
Frontend	Vanilla JavaScript (no framework), CSS Custom Properties
Database	SQLite (WAL mode, foreign keys)
Auth	bcrypt + itsdangerous (cookie sessions)
Encryption	Fernet (credentials at rest)
File Storage	Nextcloud WebDAV proxy
Documents	ONLYOFFICE via WOPI
Directory	LDAP/Active Directory (ldap3)
AI Agent Interface	MCP via FastMCP, Bearer tokens, audit log

Development

# Run in development mode
source venv/bin/activate
python dashboard/app.py        # Main app on :8504
python dashboard/admin_app.py  # Admin app on :8505

Adding a New Tab

1. Create JS file: dashboard/static/js/tab_example.js
2. Define init function: async function initExampleTab() { ... }
3. Register in app_core.js: add to switchTab() and tabMap/buildPath
4. Add to index.html: tab link in <nav class="tabs"> + <script> include
5. Optional: add API router in a new api_example.py

Components

• ExpandableTable - Sortable, filterable table with expandable detail rows
• KPI Cards - Dashboard widgets with KPI cards and sections
• WYSIWYG Editor - ContentEditable-based rich text editor with HTML sanitization

Security

• CSRF protection (X-Requested-With header validation)
• Content Security Policy (CSP) headers
• Rate limiting on login and password changes
• SQL identifier whitelisting
• HTML sanitization (XSS prevention)
• Fernet encryption for stored credentials
• MCP tokens stored as hashes; global MCP kill switch
• Audit logging for MCP and relevant write operations
• Path traversal protection for file operations
• Systemd hardening (NoNewPrivileges, ProtectSystem, PrivateTmp)

License

MIT