vector-memory-mcp
by AerionDyseti
Overview
A local-first semantic memory server for AI assistants to store and retrieve context through semantic search.
Installation
bunx @aeriondyseti/vector-memory-mcpEnvironment Variables
- VECTOR_MEMORY_DB_PATH
- VECTOR_MEMORY_MODEL
Security Notes
The `MemoryRepository` directly interpolates user-provided `id` strings into LanceDB `where` clauses (e.g., `id = '${id}'`). The MCP tool schemas define `id` as a generic `string`, without validation for UUID format or sanitization. This pattern is vulnerable to SQL-like injection attacks if a malicious MCP client provides a crafted `id` string (e.g., `' OR 1=1 --`) potentially leading to unauthorized data access, modification, or deletion within the local database. Additionally, `JSON.parse` is used on metadata retrieved from the database; while generally safe, it could pose a risk if malicious JSON is somehow stored.
Similar Servers
memorizer-v1
A .NET-based service for AI agents to store, retrieve, and search through long-term memories using vector embeddings, PostgreSQL (pgvector), and a Model Context Protocol (MCP) API, featuring versioning, relationships, and asynchronous chunking.
context-sync
Providing persistent memory and context synchronization for AI development across various tools and projects.
claude-conversation-memory-mcp
Provides long-term memory for AI coding assistants by indexing conversation history with semantic search, decision tracking, and cross-project search.
cortexgraph
CortexGraph is an MCP server that provides short-term memory for AI assistants, featuring temporal decay, reinforcement, and automatic promotion to long-term storage.