mindpm is a Model Context Protocol (MCP) server designed to provide persistent project memory for large language models (LLMs) using SQLite. This powerful tool allows LLMs to maintain context between sessions, avoiding the need to repeatedly re-explain project details in every interaction.

Key Features

• Persistent Context: Automatically track project status, decisions, architecture notes, and session history, ensuring smooth continuity across conversations.

  For example, when prompted about next steps, the LLM can instantly retrieve:

  You: "What should I work on next?"
  LLM: [queries mindpm] "Last session you finished the auth refactor. You have 3 high-priority tasks: rate limiting, API docs, and the webhook retry bug. Rate limiting is unblocked — start there."
  

• Comprehensive Tracking: mindpm records vital components of a project, including:

  • Tasks: Status, priority, blockers, and sub-tasks.
  • Decisions: What was decided, the rationale behind it, and alternative options considered.
  • Notes: Important details about architecture, bugs, ideas, and research.
  • Context: Key-value pairs like tech stack and project conventions.
  • Sessions: Detailed summaries of what has been accomplished and future steps.

• Integrated Kanban Board: A built-in Kanban UI facilitates project management. Upon starting the MCP server, users can access their board via http://localhost:3131:

  Kanban board: http://localhost:3131?project=<project-id>
  

• MCP Tools for Easy Management: mindpm comes equipped with a range of tools to manage projects effectively:

  • Projects: Create, list, and get status updates.
  • Tasks: Create, update, and filter task lists while retrieving details about individual tasks.
  • Decisions: Log decisions and maintain an easily browsable history.
  • Notes & Context: Add notes and perform full-text searches across all project-related materials.
  • Sessions: Start and end sessions seamlessly, maintaining continuity in project management.

How It Works

1. A conversation begins where the project is mentioned.
2. The LLM initiates a session, retrieving complete context through start_session.
3. Throughout the discussion, tasks are created, decisions are logged, and notes are added.
4. At the end of the session, the state is saved with end_session.
5. In subsequent conversations, the LLM is equipped with instant context, eliminating the need for re-explanation.

Setup and Environment Configuration

mindpm accommodates integration with various clients, enabling a unified JSON format for configuration. The database, which is shared across all clients, ensures that no context is lost when switching tools mid-project.