eion

Overview

Eion is a comprehensive shared memory storage solution designed to connect AI agents through unified knowledge graph capabilities. It offers a combination of memory storage using PostgreSQL with pgvector and knowledge graph functionality using Neo4j. The MCP server implementation exposes 8 tools across memory and knowledge domains, enabling agents to store and retrieve conversation memories, perform semantic searches, and manage knowledge entries. The system supports various deployment scenarios from single LLM applications to complex multi-agent systems.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• get_memory

  Retrieve conversation history with filters

  Try:Get conversation history for session 123Try:Retrieve previous messages in the current conversation
• add_memory

  Store new conversation messages

  Try:Add this message to the conversation memoryTry:Save this response to our conversation history
• search_memory

  Semantic search in conversation history

  Try:Find messages similar to 'project deadline'Try:Search for previous discussions about the UI design
• delete_memory

  Remove conversation data

  Try:Delete all messages from this conversationTry:Remove outdated information from memory
• search_knowledge

  Find relevant extracted knowledge

  Try:Search for information about company policiesTry:Find knowledge entries related to user authentication
• create_knowledge

  Add new knowledge entries

  Try:Add this fact to our knowledge baseTry:Store this information about our products
• update_knowledge

  Modify existing knowledge

  Try:Update the information about our pricing policyTry:Correct the outdated market data
• delete_knowledge

  Remove knowledge entries

  Try:Remove the outdated product informationTry:Delete this incorrect knowledge entry

Comparable tools

Installation

git clone <repo>
cd eion
docker-compose up -d

docker exec eion_postgres psql -U eion -d eion -c "CREATE EXTENSION IF NOT EXISTS vector;"
docker exec -i eion_postgres psql -U eion -d eion < database_setup.sql

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

go build -o eion-server ./cmd/eion-server
./eion-server

{
  "mcpServers": {
    "eion": {
      "command": "python",
      "args": ["-m", "internal.mcp.server"],
      "env": {"EION_BASE_URL": "http://localhost:8080"},
      "cwd": "/path/to/eion"
    }
  }
}

FAQ

What types of data can Eion store?

Eion stores conversation history in PostgreSQL with vector capabilities and extracted knowledge graphs in Neo4j. It supports semantic search, temporal knowledge storage, and provides embedding capabilities using all-MiniLM-L6-v2.

How does the MCP server integrate with external agents?

The MCP server exposes 8 tools (4 memory and 4 knowledge tools) that allow agents to store/retrieve memories, perform semantic search, and manage knowledge entries. Agents must be registered via the API and authenticate with session, agent, and user IDs.

On Hacker News

Recent discussion from the developer community.

• Show HN: Eion – a shared memory system for AI agents▲ 3💬 5
  Story by mingyk · 2025-06-20
• Eion: Shared Memory Storage for Multi-Agent Systems▲ 3
  Story by handfuloflight · 2025-06-22

Compare eion with