LM-Tasker

Overview

LM-Tasker is a straightforward task management tool for software development. It favors explicit, manual operations and provides a CLI, a terminal UI (TUI) for in-terminal task browsing and updates, and an MCP server for editor integration. Based heavily on eyaltoledano's excellent https://github.com/eyaltoledano/claude-task-master. I removed all the AI functionality because actually your usual coding assistant/agent paired with Perplexity MCP for research (or just their built-in web search tool call) has enough context to do a great job without it.

What it offers

• Manual task management: Create, organize, and track tasks and subtasks with a clear structure.
• Dependency management: Add, remove, validate, and fix task dependencies.
• Task file generation: Generate individual, human-readable files for tasks and subtasks.
• Batch operations: Update statuses, reorganize, or move tasks in bulk.
• Editor and CLI integration: Use the MCP server from supported editors (Cursor, VS Code, Windsurf) or work from the command line.
• Documentation: Practical examples and references are included in the repository.

Core capabilities

• Project initialization (auto-creates structure on first task)
• Task CRUD (create, read, update, delete)
• Status and priority tracking
• Hierarchy reorganization (move tasks/subtasks)
• Configuration and generation utilities

Requirements

LM-Tasker is a manual task management system with no external services. You will need:

• Node.js (v14 or higher)
• npm or yarn for package management

LM-Tasker works offline and doesn't require any API keys or external services.

Quick Start

Option 1: MCP (recommended)

The Model Control Protocol (MCP) lets you run LM-Tasker directly from supported editors.

1. Add your MCP config at the appropriate path for your editor

| Editor | Scope | Linux/macOS Path | Windows Path | Key | | ------------ | ------- | ------------------------------------- | ------------------------------------------------- | ------------ | | Cursor | Global | ~/.cursor/mcp.json | %USERPROFILE%\.cursor\mcp.json | mcpServers | | | Project | <project_folder>/.cursor/mcp.json | <project_folder>\.cursor\mcp.json | mcpServers | | Windsurf | Global | ~/.codeium/windsurf/mcp_config.json | %USERPROFILE%\.codeium\windsurf\mcp_config.json | mcpServers | | VS Code | Project | <project_folder>/.vscode/mcp.json | <project_folder>\.vscode\mcp.json | servers |

Cursor & Windsurf (mcpServers)

{
  "mcpServers": {
    "lm-tasker": {
      "command": "npx",
      "args": ["-y", "@qubeio/lm-tasker-mcp"],
      "env": {},
    },
  },
}

VS Code (servers + type)

{
  "servers": {
    "lm-tasker": {
      "command": "npx",
      "args": ["-y", "@qubeio/lm-tasker-mcp"],
      "env": {},
      "type": "stdio",
    },
  },
}

2. (Cursor only) Enable LM-Tasker MCP

Open Cursor Settings (Ctrl+Shift+J) → open the MCP tab → enable lm-tasker.

3. Create your first task

In your editor's AI chat pane, say:

Create a new task for implementing user authentication

This initializes the project structure and creates your first task.

4. Continue creating tasks

LM-Tasker uses manual task creation. Create tasks using the MCP tools or CLI commands.

5. Common actions in chat-enabled editors

• Create a task: Create a new task for implementing user authentication.
• Find the next task: What's the next task I should work on?
• Work on a specific task: Help me implement task 3.
• Add subtasks: Add subtasks to task 4.

More examples on how to use LM-Tasker in chat

Option 2: Command line

Installation

# Install globally
npm install -g @qubeio/lm-tasker

# OR install locally within your project
npm install @qubeio/lm-tasker

Create your first task (auto-initializes project)

# If installed globally
lm-tasker add-task --title="Task Title" --description="Task description"

# If installed locally
npx @qubeio/lm-tasker add-task --title="Task Title" --description="Task description"

This creates the tasks.json file and project structure when you add your first task.

Common Commands

# Create first task (auto-initializes project)
lm-tasker add-task --title="Task Title" --description="Task description"

# List all tasks
lm-tasker list

# Show the next task to work on
lm-tasker next

# Generate task files
lm-tasker generate

Terminal UI (TUI)

LM-Tasker includes a full-screen terminal UI for browsing tasks and viewing details.

Launch

# Start the TUI
lm-tasker ui

# Alias
lm-tasker tmui

# Specify a tasks file explicitly
lm-tasker ui --file ./tasks/tasks.json

Navigation

• Arrow keys or j/k: move selection
• Enter: open task details
• s: update task status (opens status modal)
• Esc or q: go back/exit
• h/l: navigate between list and detail views

The window title includes the current repository name to help distinguish sessions.

Screenshots

The TUI reads data directly from tasks.json. Ensure it exists before launching.

Documentation

For details, see the documentation in the docs directory:

• Configuration — Environment and customization
• Tutorial — Getting started
• Command reference — Available commands
• Task structure — Task format and features
• Examples — Common editor interaction workflows

Troubleshooting

If lm-tasker add-task doesn't respond

Try running it with Node directly:

node node_modules/lm-tasker/bin/lm-tasker.js add-task --title="Task Title" --description="Task description"

Or clone the repository and run:

cd lm-tasker
npm install
npx lm-tasker add-task --title="Task Title" --description="Task description"

Codex (or strict MCP clients) can't load the server

Some MCP clients – Codex in particular – reject any non-JSON text that appears on stdout while the server is connected over stdio. Earlier versions of LM-Tasker logged startup banners to stdout, which meant Codex aborted the handshake immediately even though Claude/Desktop would tolerate it. Upgrade to the latest version (or pull main) so logging goes to stderr only. If you maintain a fork, make sure any console.log statements are replaced with stderr logging before the MCP server starts.

Development Workflow

This repository now ships with a Taskfile.yml so common actions can be run with GoTask. Typical examples:

# Run the fast Jest suite
task test

# Run coverage
task test:coverage

# Check formatting
task lint

All tasks automatically set LMTASKER_SKIP_MCP_AUTOSTART=1 so the MCP server does not auto-run during unit tests.