MiniMe-MCP Server

MiniMe-MCP provides a persistent memory layer for your AI agents, enabling you to remember patterns, decisions, and documents across IDEs and projects. It helps your AI stay contextually aware, reducing repetitive setup explanations and enabling cross-project learning while keeping data local and private.

How to use

You connect to the MiniMe-MCP server from your MCP client in your IDE. Use the server to search memories, pull cross-project insights, and store documents or decisions so your AI agents can reason with your history. Typical workflows include building authentication patterns from memory, extracting cross-project optimizations, and applying documented guidelines to new code. You can also link related projects, upload PDFs or specs for instant understanding, and manage tasks with saved context.

How to install

Prerequisites: you need Docker and Docker Compose installed on your machine.

1. Start MiniMe-MCP server using Docker Compose.

# 1. Start MiniMe server
cd install
docker compose --env-file minime.env up -d
```,

Additional notes

The server exposes a unified MCP endpoint for clients. Use the default endpoint http://localhost:8000/mcp for interactions from your MCP-enabled IDEs and tools.

Supported IDEs include Cursor, Claude Desktop/Code, VS Code, Windsurf, JetBrains, Zed, and others that speak MCP.

Configuration and usage notes

• Memory search, cross-project insights, and document search are powered by a three-tier structure: documents, chunks, and memories. This enables fast, contextual retrieval across projects.

• Memory lifecycle management lets you update, inactivate, or reactivate memories with an audit trail to prevent drift as your stack evolves.

• Global rules can guard behavior across all agents and projects, ensuring destructive actions are confirmed and that memory searches are performed before implementing changes.

Troubleshooting tips

If the server does not respond at http://localhost:8000/mcp, verify Docker is running, check the minime.env file for correct settings, and look at docker logs for the container to identify startup errors.

Examples of what you can do with memories

• Search your memories for authentication patterns to reuse proven security approaches across projects.

• Upload API documentation or PDFs to create a searchable knowledge base that informs future code generation.

• Link related projects to share learned patterns and avoid repeating mistakes.

Security and privacy

The server is designed to run locally so your data stays on your machine. You control what is stored and how it is organized, with audit trails for changes.

Available tools

store_memory

Save decisions, patterns, learnings, and notes so your AI agents can recall them across sessions and projects.

search_memories

Perform semantic, keyword, or hybrid searches across memories and documents to surface relevant context.

expand_memories

Retrieve full content and surrounding context for memories to provide deeper insight.

modify_memory

Update, inactivate, or reactivate memories with an audit trail to maintain accuracy over time.

get_insights

Analyze cross-project patterns, identify technical debt, and surface quality signals across memories.

start_thinking

Initiate structured reasoning sequences to organize complex problem solving.

add_thought

Append intermediate reasoning steps to support transparent decision making.

manage_tasks

Create, complete, and list tasks with linked memories to track work and decisions.

get_rules

Load global and project-specific guardrails to enforce consistent behavior.

manage_project

Handle briefs, PRDs, implementation plans, and project linking to organize work.

help

Provide interactive guidance for tool selection and workflows.