Taskman

Project management for AI-assisted workflows: phases, milestones, and role-assigned tasks with dependencies, stored in SQLite. Agents (via MCP) create and update work; humans interact through Cursor or other MCP clients. The CLI is for scripts and automation, the web UI for read-only browsing; output is JSON or plain text. It runs as a standalone C++ binary on Windows, Linux, and macOS.

Quick start

1. Download taskman — Latest release (Windows, Linux, macOS).
2. Bootstrap the project — From your project root, run taskman by its full path:

   /path/to/taskman project:init
   

   This writes .cursor/mcp.json, creates the database, and generates .cursor/rules/ and .cursor/agents/.
3. Reload Cursor — So the MCP server is loaded.
4. Use Taskman via the agent — Ask the agent to manage phases, milestones, and tasks.

Documentation

• Human users via agents: who uses Taskman and which agent roles
• Usage overview: data model, interfaces, workflows, task assignment
• MCP Server: integration with AI assistants
• CLI Usage: command-line interface, commands, options, formats
• Web Interface: browser-based UI (read-only - edit planned)
• Build: download, build and run instructions

Changes

• User changelog (non-developer changes)
• Global Changelog (all changes)