a-mem-mcp-server
Verified Safeby tobs-code
Overview
An agentic memory system for LLM agents that stores, links, evolves, and retrieves information based on the Zettelkasten principle, enhanced with autonomous maintenance enzymes and deep web research capabilities.
Installation
python -m src.a_mem.mainEnvironment Variables
- LLM_PROVIDER
- OLLAMA_BASE_URL
- OLLAMA_LLM_MODEL
- OLLAMA_EMBEDDING_MODEL
- OPENROUTER_API_KEY
- OPENROUTER_BASE_URL
- OPENROUTER_LLM_MODEL
- OPENROUTER_EMBEDDING_MODEL
- MAX_NEIGHBORS
- MIN_SIMILARITY_SCORE
- TCP_SERVER_ENABLED
- TCP_SERVER_HOST
- TCP_SERVER_PORT
- RESEARCHER_ENABLED
- RESEARCHER_CONFIDENCE_THRESHOLD
- RESEARCHER_MAX_SOURCES
- RESEARCHER_MAX_CONTENT_LENGTH
- JINA_READER_ENABLED
- JINA_READER_HOST
- JINA_READER_PORT
- UNSTRUCTURED_ENABLED
- UNSTRUCTURED_API_URL
- UNSTRUCTURED_API_KEY
- UNSTRUCTURED_USE_LIBRARY
- GOOGLE_SEARCH_ENABLED
- GOOGLE_API_KEY
- GOOGLE_SEARCH_ENGINE_ID
- GRAPH_BACKEND
- REDIS_HOST
- REDIS_PORT
- REDIS_PASSWORD
Security Notes
The system externalizes API keys and configurations via .env files. It implements input validation for MCP tool parameters, reducing common injection risks. The optional HTTP server for graph visualization is read-only. External network calls (Ollama, OpenRouter, Google Search, Jina Reader, Unstructured) are handled using the 'requests' library, relying on standard API security. However, the Researcher Agent's web content fetching and processing of external data sources introduces potential attack surfaces (e.g., if URLs are not perfectly sanitized before fetching, or if fetched content could lead to processing vulnerabilities, though current implementations use robust libraries like Jina Reader and Unstructured). Local file I/O for data storage (graph, chroma, events) requires the host system to have appropriate file permissions.
Similar Servers
memory-graph
A graph-based MCP server that provides intelligent memory capabilities for Claude Code, enabling persistent knowledge tracking, relationship mapping, and contextual development assistance.
memory-mcp-server-go
A Model Context Protocol server providing knowledge graph management capabilities for LLMs to maintain memory across conversations.
nocturne_memory
Provides a persistent, structured long-term memory and dynamic knowledge graph system for AI agents, designed for human-AI collaboration.
kuzu-memory
KuzuMemory provides a lightweight, embedded graph-based memory system for AI applications, specifically designed to integrate with AI coding agents (like Claude Code) via the Model Context Protocol (MCP) for contextual prompt enhancement and learning from interactions.