Overview

MemoryLayer is an API-first memory infrastructure for LLM-powered agents. It solves the fundamental challenge of stateless LLMs by providing persistent, queryable, multi-modal memory that agents can read from and write to during execution.

What is MemoryLayer?

MemoryLayer provides cognitive memory capabilities for AI agents, including episodic, semantic, procedural, and working memory with vector-based retrieval and graph-based associations.

Think of it as a memory backend for your AI applications — the same way you’d use PostgreSQL for relational data or Redis for caching, you use MemoryLayer for agent memory.

Key Features

• Cognitive Memory Architecture — Episodic, semantic, procedural, and working memory types modeled after human cognition
• Vector Search — SQLite + sqlite-vec for efficient similarity search; cloud embeddings (OpenAI / Google) or self-hosted text & multi-vector via the memorylayer-embed-server peer
• Knowledge Graph — 63 typed relationships across 11 categories (causal, solution, learning, workflow, hierarchical, etc.) connecting memories into a traversable semantic graph
• Context Environment — Server-side Python sandbox with RLM (Recursive Language Model) so heavy memory analysis doesn’t consume the agent’s context window
• Session Management — Working memory with TTL, token-budget-aware extraction triggers, and commit to long-term storage
• Multi-Platform SDKs — Python and TypeScript client libraries with full type safety
• Framework Integrations — Drop-in support for LangChain, LlamaIndex, Claude Code (plugin + MCP), and OpenCode (plugin + MCP)
• MCP Server — 25 tools in the default profile (38 in full) for Claude Code, Claude Desktop, Cursor, and other MCP-compatible LLMs
• REST API — FastAPI-based HTTP server with OpenAPI documentation

Architecture Overview

┌─────────────────────────────────────────────────────────────┐
│  Agent Frameworks (LangChain, LlamaIndex, Claude Code)      │
├─────────────────────────────────────────────────────────────┤
│  Client SDKs (Python, TypeScript)                           │
├─────────────────────────────────────────────────────────────┤
│  MemoryLayer Server (REST API + MCP)                        │
├─────────────────────────────────────────────────────────────┤
│  Storage (SQLite + sqlite-vec)                              │
└─────────────────────────────────────────────────────────────┘

Packages

Package	Install	Description
Server	pip install memorylayer-server	FastAPI server with SQLite storage
Python SDK	pip install memorylayer-client	Async Python client library
TypeScript SDK	npm install @scitrera/memorylayer-sdk	TypeScript/JavaScript client
MCP Server	npm install @scitrera/memorylayer-mcp-server	Model Context Protocol integration
LangChain	pip install memorylayer-langchain	LangChain memory backend
LlamaIndex	pip install memorylayer-llamaindex	LlamaIndex chat store

Quick Example

from memorylayer import MemoryLayerClient, MemoryType

async with MemoryLayerClient(base_url="http://localhost:61001") as client:
    # Store a memory
    memory = await client.remember(
        content="User prefers Python for backend development",
        type=MemoryType.SEMANTIC,
        importance=0.8,
        tags=["preferences", "programming"]
    )

    # Recall memories
    results = await client.recall(
        query="What programming languages does the user like?",
        limit=5
    )

    # Synthesize insights
    reflection = await client.reflect(
        query="Summarize user's technology preferences"
    )
    print(reflection.reflection)

Next Steps

• Installation — Install the server and your preferred SDK
• Core Concepts — Understand memory types, workspaces, and associations
• Python SDK Quick Start — Build your first memory-powered application
• TypeScript SDK Quick Start — Get started with TypeScript