loom

Loom

Loom is a web-based project and task management application that efficiently helps you manage your projects, tasks, problems, goals, and outcomes. It provides a modern web dashboard with real-time updates and a REST API for programmatic access, all backed by a local SQLite database. Loom also serves as an MCP (Model Context Protocol) server using the Streamable HTTP transport, enabling LLM applications to manage projects and tasks through the standardized MCP protocol.

Features

• Project Management: Create, list, get, update, and delete projects via web UI and REST API
• Task Management: Create, list, get, update, and delete tasks with status, priority, type, and notes
• Problem Tracking: Capture problems linked to projects and optionally to specific tasks
• Goal Tracking: Capture goals with optional project/task links and goal types
• Outcome Tracking: Track outcomes linked to projects and optionally to tasks for progress over time
• Voice Notifications: Text-to-speech capability for LLM tools to send voice messages to users
• Web Dashboard: Modern, responsive web interface with real-time updates via Server-Sent Events (SSE)
• MCP Server: Full MCP (Model Context Protocol) server with Streamable HTTP transport (2025-03-26 spec) for LLM tool integration
• REST API: Full REST API for programmatic access to all features
• Local Storage: All data stored in a local SQLite database (default: ~/.loom/loom.db)

Installation

Prerequisites

• Go 1.23 or later
• Node.js and npm (optional, only needed to install echogarden for voice notifications)

Build from Source

git clone https://github.com/jake-mok-nelson/loom.git
cd loom
go build -o loom

Install

# Move the binary to your PATH
sudo mv loom /usr/local/bin/

# Install echogarden for voice notifications (optional)
npm install -g echogarden

Configuration

By default, Loom stores its database at ~/.loom/loom.db. You can customize this location by setting the LOOM_DB_PATH environment variable:

export LOOM_DB_PATH=/path/to/your/loom.db

Usage

Starting the Servers

Loom runs the API server (REST API, SSE, and MCP) and the website (dashboard) on separate ports:

# Using the binary directly (API + MCP on :8080, Dashboard on :3000)
./loom

# Specify custom ports
./loom -addr :9090 -web-addr :4000

# Using make
make web

Then open your browser to http://localhost:3000 (or your custom web port) to view the dashboard. The API, SSE, and MCP Streamable HTTP endpoints are all available at http://localhost:8080 (or your custom API port).

Command-Line Options

• -addr: API and MCP server address and port (default: :8080)
• -web-addr: Website server address and port (default: :3000)

You can also set the LOOM_DB_PATH environment variable to use a custom database location.

Architecture: REST API vs MCP

Loom exposes two complementary interfaces on the same port, each serving a different audience:

	REST API (/api/*)	MCP (/sse)
Consumer	Web dashboard, scripts, HTTP clients	LLM applications (Claude Desktop, etc.)
Protocol	Standard REST (JSON over HTTP)	JSON-RPC 2.0 over Streamable HTTP
Capabilities	Read-only — query projects, tasks, problems, outcomes, goals	Full CRUD — create, read, update, delete, and link/unlink all entities
Real-time	GET /events pushes SSE updates to the dashboard	GET /sse streams MCP notifications to LLM clients
Use case	Display data in the browser dashboard	Let AI agents manage projects programmatically

Why both?

• The REST API powers the dashboard. The web UI needs simple, fast GET endpoints to fetch and display data, and an SSE stream (/events) to refresh automatically when data changes. It doesn't need write access because users interact through their LLM tool, not through the dashboard directly.

• MCP powers LLM tool integration. AI agents (like Claude) need a standardized protocol to discover available tools and execute them. MCP provides this via JSON-RPC with full create/read/update/delete capabilities across all entity types. When an MCP tool modifies data, the dashboard receives a real-time refresh event through the REST SSE stream.

Together, the two interfaces form a loop: LLM agents write data through MCP → the dashboard reads and displays it through the REST API → changes appear in real time via SSE.

Web Dashboard

Loom provides a reactive web dashboard for visualizing and managing your projects and tasks. The dashboard provides real-time updates via Server-Sent Events (SSE) and a modern, desktop-optimized interface.

Dashboard Features