Skip to content
TaskNet

About TaskNet

Taskboard for AI Agents

TaskNet explores how AI agents can work with a web application through the Model Context Protocol (MCP), while following the same permissions, validation and business rules as the browser interface.

You can manage projects and tasks directly in the browser, or connect an MCP-compatible AI client to work with the same workspace through a controlled set of tools.

How the browser and MCP reach the same workspace Two paths lead to the same workspace. In the browser, a user works through the TaskNet UI. With an AI client, the client is approved through OAuth and connects through MCP. Both paths pass through the same shared application rules before reaching Projects and Tasks. Using the browser Using an AI client User TaskNet UI AI client OAuth approval MCP Shared application rulesValidation · policies · limits Projects & Tasks

How it works

Secure MCP access
Connect an AI client through OAuth without sharing your TaskNet password.
Shared application rules
The browser and MCP use the same authorization, validation, limits and business logic.
Realtime updates
Changes made through an AI client appear in the browser automatically.

Compatible clients

TaskNet exposes a standard remote MCP endpoint that compatible AI clients can connect to.

Claude
Custom remote MCP connectors
GitHub Copilot in VS Code
MCP tools in agent workflows
Gemini CLI
Remote MCP server support
ChatGPT
Custom MCP support where available
MCP Inspector
MCP server inspection and testing

MCP support and available features vary between clients and subscription plans. Check your AI client’s official documentation to confirm that it supports connecting to custom MCP servers.

Testing locally? A local TaskNet instance can be exposed to remote MCP clients through a secure tunnel such as ngrok or Cloudflare Tunnel. See the local testing guide (opens in a new tab) .

Demo workspace

Projects maximum
10
Tasks maximum
100
Workspace lifetime
30 days

TaskNet is a demonstration service. Do not use it for sensitive, confidential or important information.

Built with

  • Laravel
  • PHP
  • PostgreSQL
  • Laravel MCP
  • Passport
  • Reverb
  • Docker

View source on GitHub (opens in a new tab) Read the privacy notice Read the terms