ATLAS: Task Management System

What is ATLAS: Task Management System?

ATLAS (Adaptive Task & Logic Automation System) is a project, knowledge, and task management system for LLM agents, built on a 3-node architecture of Projects, Tasks, and Knowledge. It uses Neo4j as its graph database and communicates via the Model Context Protocol (MCP), enabling standardized interaction between AI models and external systems.

How to use ATLAS: Task Management System?

Clone the repository, install dependencies (npm install), configure a Neo4j instance (self-hosted via Docker or cloud via Neo4j AuraDB), build the project (npm run build), and run the server using npm run start:stdio (default) or npm run start:http. Environment variables such as NEO4J_URI, NEO4J_USER, and NEO4J_PASSWORD must be set in your MCP client config or a .env file.

Key features of ATLAS: Task Management System

• Project management with metadata, statuses, and bulk operations
• Task lifecycle management with prioritization and categorization
• Knowledge repository with domain categorization and citation support
• Native dependency tracking for projects and tasks
• Cross-entity search with fuzzy matching and wildcard support
• Multiple transport modes: stdio and Streamable HTTP

Use cases of ATLAS: Task Management System

• LLM agents tracking and updating project progress autonomously
• Managing complex task workflows with prerequisite dependencies
• Building and querying a structured knowledge base for project teams
• Unified search across projects, tasks, and knowledge items
• Integrating task management into IDEs or chat-based assistants via MCP

FAQ from ATLAS: Task Management System

What database does ATLAS use?

ATLAS uses Neo4j (v2.0+). Earlier versions (v1.5.4 and before) used SQLite. Neo4j can be self-hosted with Docker (docker-compose provided) or via Neo4j AuraDB cloud service.

How do I run ATLAS with different transport modes?

Set the environment variable MCP_TRANSPORT_TYPE to stdio (default) or http. For HTTP mode, also configure MCP_HTTP_HOST, MCP_HTTP_PORT, and optionally MCP_ALLOWED_ORIGINS for CORS.

Is there a web UI?

Yes, an experimental Web UI is available. Run npm run webui to open it in your browser for viewing Project, Task, and Knowledge details.

How is data secured for HTTP transport?

If using HTTP transport, you can enable JWT authentication by setting MCP_AUTH_SECRET_KEY (minimum 32 characters). Rate limiting is also available via MCP_RATE_LIMIT_WINDOW_MS and MCP_RATE_LIMIT_MAX_REQUESTS (default: 100 requests per minute per IP). Production use of HTTP transport has not been fully tested.

Can I back up the database?

Yes, ATLAS includes backup and restore commands. Backups are stored in the directory specified by BACKUP_FILE_DIR (default ./atlas-backups), and you can set BACKUP_MAX_COUNT to limit the number of backup sets kept.