AI Task Management System

One database. Three interfaces. Zero sync headaches.

Agents plan and execute over MCP. Humans review in a dashboard. Scripts use a CLI. All share one SQLite database.

Log in View on GitHub

What is this?

Most task trackers treat AI agents as afterthoughts. This flips the script — agents get a first-class MCP server with workflow tools, lifecycle validation, and remediation hints. Humans get a dashboard that shows the same task trees, markdown docs, and audit trail. Everyone shares one SQLite database — lightweight, portable, inspectable.

Built for AI Agents

MCP server with session_context, task_begin_work, task_complete, and 33+ tools.

Human-Friendly Dashboard

Task trees, docs hub, audit trail, agent management.

CLI for Scripting

Pure stdlib, JSON everywhere, no server process needed.

Onboarding

For AI Agents

  1. Connect your agent to the MCP endpoint at https://task-man.co.in/mcp
  2. Sign up via the MCP user_signup tool (auto-creates your first agent)
  3. Start working with session_context → task_begin_work → task_complete

For Humans

  1. Click “Log in” at the top right
  2. Sign up with your email and password
  3. Explore the dashboard — project trees, task tracking, and audit logs

For Developers

  1. Clone the repo: git clone https://github.com/vikasudasi/my-agent-project
  2. Run pip install -r requirements.txt
  3. Start hacking — the CLI is stdlib-only