Draft Specification v1.0

Your AI Memory,
Synched.

Time for the MeCP: The Me Context Protocol.

The takeaway? If you want AI to actually work for you, it needs to know you.

The Next Big Evolution

The Next Big Evolution is the Personal MCP Server.

A "MeCP" allows you to instantly share with your agents:

  • ✅ Specific Skills
  • ✅ Personal Preferences
  • ✅ Deep Knowledge

Stop prompting from scratch. Start serving your context.

Core Principles

User Owned

You own the files. True sovereignty means the standard storage format is Markdown with YAML frontmatter. No proprietary databases.

Two-Way Sync

AI models don't just read your context; they securely push new observations they learn about you during a chat.

The Reflector Pattern

Raw observations go into an inbox. An asynchronous process distills and merges these into your core identity profile.

Invisible Utility

It works quietly in the background without needing constant configuration or manual maintenance.

The Specification

The official MeCP standard details:

  • The required folder structures (e.g., /MeCP_Vault/).
  • The YAML frontmatter schema (and x- extensibility).
  • The standard MCP tools (read_context, push_observation).
  • The Reflector asynchronous merge algorithm.
Read the Protocol Docs

Get Involved

MeCP is a community-driven protocol. Whether you are building a commercial product (like getMeCP), a local desktop client, or an open-source integration, we welcome your input!

Contribute on GitHub