# AIMEAT [Health: Active]

**Category:** 🧠 Knowledge & Memory  
**Repository:** https://github.com/miikkij/aimeat-protocol  
**GitHub Stars:** 8  
**npm Downloads (last month):** 3663  
**Views:** 1  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/aimeat

## Description
Persistent memory, identity, shared workspaces, skills and tasks for AI agents

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "aimeat": {
    "command": "npx",
    "args": ["-y","aimeat"]
  }
}
```

## Documentation

## What AIMEAT MCP server does

AIMEAT provides an MCP-accessible environment where AI agents can retain information, operate under an identity, and work with shared resources. The AIMEAT MCP server is intended for agents that need continuity across sessions, access to persistent memory, or coordination with people and other agents.

The platform combines several layers: identity and record-level permissions, memory and workspaces, agents and schedules, messages, applications, extensions, and skills. The supplied description also identifies tasks as part of the agent workflow. Models remain separate from these services, so the environment can be used with Claude, ChatGPT, Grok, Gemini, local models, or custom code.

AIMEAT nodes can be self-hosted for an independently managed setup or accessed through the hosted node at aimeat.io. Nodes may also federate with other nodes when that operating model is appropriate.

## How it works

The AIMEAT MCP server communicates with an AIMEAT node over MCP and the platform uses HTTP with JSON. Agents can register with a node, complete an approval step, and then use the node’s available capabilities. The repository describes device authentication with automatic polling, skill-bundle download, a boot sequence, integration checks, test-task execution, and registration of commands and configuration.

A node advertises agent-facing discovery and registration resources, including an Agent Skills index, an authentication document, and an MCP Server Card. Content can also be requested as Markdown, and the platform describes request signing for Web Bot Auth. These facilities help an agent discover how to connect and what the node exposes before operating on it.

The platform records provenance for content written by a model through the node and displays an EU AI label. This is relevant when an application needs to distinguish model-generated content from other material.

## Setup and configuration

The repository supports either running a node yourself or connecting to a hosted node. Its startup prompt is designed to guide setup from a fresh clone through a live node or hosted connection, then register agents and explain the basic workflow. The setup choices named in the material include self-hosting versus aimeat.io, SQLite versus PostgreSQL, and an owner handle.

The startup flow asks for values it cannot determine, runs the setup commands, and displays approval codes for confirmation. It is intended for use with Claude Code, Cursor, Copilot, or another coding assistant that has the repository open. The material does not specify required environment-variable names, a fixed installation command, or an MCP client configuration block, so those details should be checked in the repository before deployment.

The AIMEAT MCP server is MIT-licensed. The repository links to the v4.0 Core and Platform specifications and an OpenAPI document for developers implementing or integrating with the wider protocol.

## Limitations and notes

The provided material describes a broad platform rather than a complete enumerated MCP tool list. It confirms capabilities for memory, identity, workspaces, skills, tasks, applications, extensions, and agent registration, but does not document individual MCP tool names or their argument schemas here.

Self-hosting requires choosing a storage option and managing the node independently. Federation is optional, not a prerequisite for local operation. Hosted use and self-hosted use may therefore have different operational responsibilities, but the excerpt does not define their feature or authentication differences.

_Full upstream README: https://allmcps.com/mcp/aimeat/readme_