MemoraEU

pquattro/memoraeu-memoraeu-mcp

This gives Claude encrypted persistent memory with client-side AES-256-GCM encryption and semantic search. It exposes tools like `remember`, `recall`, `forget`, and `list_memories` that work automatically: Claude recalls context at session start and stores facts without prompting. Embeddings are generated locally via Mistral before encryption, so the EU-hosted server never sees plaintext. It handles deduplication by vector similarity and compresses long memories before storage. You can run it locally via stdio for zero-knowledge guarantees or connect remotely via SSE (which processes plaintext server-side). Useful when you want Claude to maintain context across conversations without exposing data to the hosting infrastructure.

tools: 10
auth: Secret needed
transport: STDIO
env vars: 5
last updated: Apr 19, 2026

View on GitHub

Tools

10 tools

Public tool metadata for what this MCP can expose to an agent.

memory.storeStore a memory with optional category and tags. Use automatically when the user shares preferences, decisions, biographical facts, or durable constraints.3 params

Store a memory with optional category and tags. Use automatically when the user shares preferences, decisions, biographical facts, or durable constraints.

Parameters* required

categorystring

Optional category label (e.g. 'preferences', 'work', 'health').

contentstring

The memory text to store.

tagsarray

Optional list of keyword tags.

memory.recallSemantic search through memories in natural language. Call automatically on the first message of each conversation.3 params

Semantic search through memories in natural language. Call automatically on the first message of each conversation.

Parameters* required

categorystring

Optional category filter to restrict results.

limitinteger

Maximum number of results to return (default 3).default: 3

querystring

Natural language query to search memories.

memory.deletePermanently delete a memory by its ID.1 params

Permanently delete a memory by its ID.

Parameters* required

memory_idstring

The full or partial ID of the memory to delete.

memory.listList recent memories with optional category filter.2 params

List recent memories with optional category filter.

Parameters* required

categorystring

Optional category to filter by.

limitinteger

Maximum number of memories to return (default 20).default: 20

memory.categoriesReturn all existing memory categories sorted by usage count.

Return all existing memory categories sorted by usage count.

No parameter schema in public metadata yet.

fact.storeStore a structured fact as subject/predicate/object with optional temporal validity. Automatically invalidates any previous fact with the same subject and predicate.5 params

Store a structured fact as subject/predicate/object with optional temporal validity. Automatically invalidates any previous fact with the same subject and predicate.

Parameters* required

objectstring

The value of the fact (e.g. 'Paris', 'engineer', 'dark mode').

predicatestring

The relationship or attribute (e.g. 'lives_in', 'role', 'prefers').

scopestring

Visibility: 'private' (user only) or 'org' (shared across organization).one of private · orgdefault: private

subjectstring

The entity the fact is about (e.g. 'user', 'project:alpha').

valid_fromstring

ISO date when this fact becomes valid (YYYY-MM-DD). Defaults to today.

fact.listRetrieve currently active facts for a subject, with optional history of past values.4 params

Retrieve currently active facts for a subject, with optional history of past values.

Parameters* required

historyboolean

If true, include invalidated past facts in addition to current ones.default: false

predicatestring

Optional predicate filter to narrow results.

scopestring

Visibility scope to query.one of private · orgdefault: private

subjectstring

The entity to retrieve facts for.

fact.invalidateMark a specific fact as expired by setting its end date.2 params

Mark a specific fact as expired by setting its end date.

Parameters* required

fact_idstring

The ID of the fact to invalidate.

valid_tostring

ISO date when the fact expired (YYYY-MM-DD). Defaults to today.

memory.searchSearch memories by natural language query and return results with relevance scores. Used by ChatGPT Deep Research mode.2 params

Search memories by natural language query and return results with relevance scores. Used by ChatGPT Deep Research mode.

Parameters* required

limitinteger

Maximum number of results to return (default 10).default: 10

querystring

Natural language search query.

memory.fetchFetch the full content of a specific memory by its ID. Used by ChatGPT Deep Research mode to retrieve complete memory details.1 params

Fetch the full content of a specific memory by its ID. Used by ChatGPT Deep Research mode to retrieve complete memory details.

Parameters* required

idstring

Full or partial memory ID to retrieve.

Configuration

MEMORAEU_API_URL

MemoraEU API endpoint (default: https://api.memoraeu.isquarecube.fr)

MEMORAEU_API_KEY*secret

Your MemoraEU API key (starts with meu-sk-). Create one at app.memoraeu.isquarecube.fr.

MEMORAEU_SECRET*secret

Your encryption secret (never sent to the server). Found in Settings on app.memoraeu.isquarecube.fr.

MEMORAEU_SALT*secret

Your encryption salt. Found in Settings on app.memoraeu.isquarecube.fr.

MISTRAL_API_KEY*secret

Mistral API key for semantic embeddings. Get one at console.mistral.ai.

Readme

memoraeu-mcp

🇬🇧 English | 🇫🇷 Français

English

Zero-knowledge persistent memory layer for Claude — MCP server

MemoraEU gives Claude a persistent, encrypted memory. All content is encrypted client-side with AES-256-GCM before reaching the server — the server never sees your plaintext. Semantic search is powered by Mistral embeddings generated locally on your machine, before encryption. (Local stdio install. The remote SSE mode is not zero-knowledge: see the note below.)

Features

🔒 Zero-knowledge — AES-256-GCM encryption, key never leaves your machine
🧠 Semantic search — embeddings generated locally before encryption (Qdrant on server)
🔄 Auto memory — remembers and recalls context automatically
🚫 Deduplication — detects near-duplicate memories before storing
🇪🇺 EU hosted — GDPR compliant infrastructure

Installation

pip install memoraeu-mcp

Or with uvx (no install required, recommended):

uvx memoraeu-mcp

Claude Desktop configuration

Add to your claude_desktop_config.json, under mcpServers:

View the full README on GitHub

Details

Repository: github.com/pquattro/memoraeu-memoraeu-mcp
Categories: AI & LLM Tools, Documents & Knowledge, Search & Web Crawling
Package: memoraeu-mcp
Transport: STDIO
Auth: Required
Registry: MCP registry, active

Related AI & LLM Tools MCP servers

View all

VEO 3.1u2n4/gemini-media-mcp4K video generation via Google VEO 3.1: text-to-video, image-to-video, video extension0stars

Model Council Mcp Serverivanantigravity-lgtm/model-council-mcp-serverThree-model MCP server returning Grok, Gemini, and DeepSeek outputs for Claude.1stars

Apple Mail MCPjayvee6/apple-mail-mcpLLM access to Apple Mail and iCloud via AppleScript. Read, search, compose, reply, flag and delete.1stars

Doctree Mcpjoesaby/doctree-mcpBM25 search + tree navigation over markdown docs for AI agents. No embeddings, no LLM calls.1stars

memexkioie/memexLocal MCP memory for coding agents. SQLite on your machine — preferences persist across chats.1stars

fish-bridge — Session Knowledge Graphmakeamouse/fish-bridge-mcpCompresses AI chat sessions into a typed knowledge graph. Local/offline via Ollama.1stars