Open Source · MIT Licensed · 100% Local

Now Your AI Will Remember Everything
& Stop Hallucinating.

Persistent memory for Claude, Cursor, Windsurf, Codex, and any MCP client. Runs locally. Zero cloud dependencies. Shadow-Decay ensures your AI never hallucinates based on stale decisions.

Get Started → View on GitHub
copy

$ pip install git+https://github.com/TimoLabsAI/ember-mcp.git && ember-mcp

# That's it. Your AI now has permanent memory.

Tired of Re-Explaining Your Stack?

Every new chat window starts from zero. You explain your tech stack, your architecture decisions, your naming conventions — again. Ember remembers everything across sessions, clients, and projects — so your chats never start from zero.

Getting Suggestions for Code You Abandoned?

You migrated from JWT to OAuth last week, but your AI still suggests JWT patterns. Shadow-Decay detects the evolution and automatically deprioritizes outdated memories.

Context Doesn't Follow You?

Debug in Claude Desktop during the day, refactor in Cursor at night — but each tool starts fresh. Ember shares memory across all MCP clients, so you pick up exactly where you left off.

Key Features

Everything you need for AI that actually remembers.

Cross-Session Memory

Close your laptop Friday, open a fresh chat Monday — the AI picks up exactly where you left off. No summaries needed.

100% Private & Local

Runs on CPU (~500MB disk, ~200MB RAM). No API keys, no cloud. Paste NDAs, proprietary code — nothing leaves your machine.

Shadow-Decay

Newer memories automatically shadow older similar ones. HESTIA scoring penalizes stale content via (1−Φ)γ, so outdated info drops in rankings without being deleted.

Temporal Intelligence

Memories ranked by recency and access frequency with importance-based half-lives. Upgrade from React 17 to 19? It stops suggesting the old syntax.

Knowledge Graph

Semantic edges connect related memories. Graph search traverses connections via BFS to discover correlated context across topics.

Source Linking

Every memory traces back to its origin. When the AI says “we chose Kubernetes,” it points to the exact doc where that was recorded.

Why Shadow-Decay Matters

Standard vector stores suffer from “semantic collision” — old memories match new queries because they’re semantically similar. Ember’s Shadow-Decay framework detects the shift and penalizes stale content.

January

You tell Claude your project uses PostgreSQL.

Ember stores memories about schemas and SQL drivers in the “databases” region of the vector space.

April

You migrate to MongoDB.

New memories about documents and collections flow into the same region. Shadow-Decay begins.

Without Ember

The old “we use PostgreSQL” memory has high similarity to “database queries.” Claude confidently gives you SQL syntax for a database you no longer use.

Hallucination based on stale memory.

With Ember

Newer MongoDB memories shadow the older PostgreSQL ones. Shadow load (Φ) approaches 1.0. HESTIA scoring applies (1−Φ)γ penalty, ranking PostgreSQL 10x lower.

Claude retrieves only MongoDB context.

How It Works

You just talk to your AI normally. Behind the scenes, Ember clusters your conversations by topic, tracks freshness through Shadow-Decay, and injects only relevant context.

1

Local Embeddings

all-MiniLM-L6-v2 generates 384-dim vectors on CPU. No API calls.

2

Semantic Clustering

16 frozen centroids cluster knowledge by topic automatically.

3

FAISS Search

Meta's FAISS handles high-speed similarity search with custom ID mapping.

4

Shadow-Decay

Newer memories shadow older similar ones. HESTIA scoring: S = cos · (1−Φ)γ · vitality.

5

Knowledge Graph

Semantic edges connect related embers. BFS traversal discovers correlated context.

6

Drift Detection

Per-cell statistics with Welford’s algorithm and Laplacian smoothing auto-flag stale regions.

Quick Start

One command. No config files to edit.

Enter your email to get the install command

We'll follow up to hear how Ember is working for you — no spam, ever.

Manual setup for unsupported clients — see docs.

16 Tools for Your AI

Exposed via the Model Context Protocol to any connected LLM.

Store & Recall

ember_store

Save a named memory with importance level and tags

ember_recall

Semantic search ranked by HESTIA score

ember_deep_recall

Recall + read source files for full context

ember_learn

Auto-capture facts, preferences, decisions

Management

ember_list

List all memories with pagination, filter by tag

ember_delete

Remove a memory by ID

ember_contradict

Mark stale, store corrected version

ember_read

Read full content of a specific memory

Intelligence

ember_auto

Auto-retrieve context at conversation start

ember_inspect

View Voronoi cell distribution and density

ember_save_session

Save session summary with source linking

ember_drift_check

Analyze region health, flag stale memories

Advanced

ember_graph_search

Vector search + BFS via knowledge graph

ember_health

Hallucination risk score with trend

ember_recompute_shadows

Recalculate all shadow loads

ember_explain

HESTIA score breakdown for any memory

Frequently Asked Questions

What is Ember MCP?
Ember MCP is a Model Context Protocol server that gives your AI long-term memory with Shadow-Decay intelligence. It stores, retrieves, and manages knowledge across all your conversations and AI tools — locally on your machine with zero cloud dependencies. The Shadow-Decay framework ensures stale information is automatically deprioritized.
Which AI tools does it work with?
Any MCP-compatible client: Claude Desktop, Claude Code, Cursor, Windsurf, and more. The ember-mcp init command auto-detects and registers with all installed clients.
Does my data leave my machine?
No. Everything runs locally — embeddings are generated on CPU using all-MiniLM-L6-v2, and all data is stored in ~/.ember/ on your filesystem. No API keys, no cloud services, no external calls.
What is Shadow-Decay?
Shadow-Decay is Ember's framework for managing knowledge freshness. When you store a new memory that's semantically similar to an older one, the newer memory “shadows” the older one — assigning it a shadow load (Φ) between 0 and 1. HESTIA scoring then penalizes shadowed memories by (1−Φ)γ, naturally pushing stale information down in retrieval rankings without deleting it.
What is HESTIA scoring?
HESTIA is Ember's retrieval ranking formula: S = cos_sim × (1−Φ)γ × vitality_factor. It combines semantic similarity with shadow load and topic vitality to ensure you get the most relevant, fresh, and actively-discussed memories — not just the most similar ones.
How much disk space and RAM does it need?
About 500MB disk (mostly the embedding model) and ~200MB RAM. It runs entirely on CPU — no GPU required.
How do I install it?
Two commands: pip install git+https://github.com/TimoLabsAI/ember-mcp.git then ember-mcp init. The init command auto-detects clients, registers Ember, creates storage, and downloads the embedding model.
Do I need to manually manage memories?
No. Ember automatically captures important information from your conversations (decisions, preferences, facts) and manages staleness through Shadow-Decay. You just talk to your AI normally.
Can I use it across multiple projects?
Yes. Ember memory is global — it works across all projects, sessions, and MCP clients. You can tag memories by project for organization.
Is it open source?
Yes. MIT licensed. View the source at github.com/TimoLabsAI/ember-mcp.