CheMiguel23/MemoryMesh
Free增强的基于图的记忆，专注于 AI 角色扮演和故事生成。
FreeFree tier
About CheMiguel23/MemoryMesh
MemoryMesh is an open-source knowledge graph server designed for AI models, with a focus on text-based RPGs and interactive storytelling. It uses the Model Context Protocol (MCP) to provide structured memory persistence across conversations, enabling AI to maintain consistent context. Based on the Knowledge Graph Memory Server from the MCP servers repository, MemoryMesh adds dynamic schema-based tools, automatic tool generation, relationship handling, an event system, and informative error feedback. Compatible with clients like Claude Desktop, ChatGPT, Cursor, Gemini, and VS Code, it runs locally and is highly customizable through user-defined schemas.
Key Features
Dynamic schema-based tools automatically generated from user-defined schemas
Relationship handling with edges between nodes (entities)
Event system tracking operations on the knowledge graph
Informative error feedback to help AI learn and improve
Local deployment with MCP compatibility for multiple AI clients
v0.3.0 update: MCP SDK updated to v1.25.2 for latest protocol compliance
Pros & Cons
Pros
- Open source and free to use
- Helps AI maintain consistent memory across conversations
- Highly customizable via schemas
- Runs locally with no external dependencies
- Supports multiple MCP-compatible clients (Claude Desktop, ChatGPT, Cursor, etc.)
Cons
- Requires Node.js and npm for installation
- Updates may require schema relocation (e.g., v0.3.0 changed default schema path)
- Limited documentation beyond GitHub README
Best For
Text-based RPGs and interactive storytellingSocial network simulationsOrganizational planningAny scenario requiring structured AI memory persistence
FAQ
What is MemoryMesh?
MemoryMesh is a local knowledge graph server that uses the Model Context Protocol (MCP) to provide structured memory persistence for AI models, particularly for text-based RPGs and interactive storytelling.
What are the key features of MemoryMesh?
Key features include dynamic schema-based tools that auto-generate from user schemas, relationship handling with nodes and edges, an event system, informative error feedback, and compatibility with MCP clients like Claude Desktop, ChatGPT, and Cursor.
What is new in v0.3.0?
The MCP SDK was updated from v1.0.4 to v1.25.2 to comply with the November 2025 MCP specification. The default schema location changed to dist/data/schemas, so existing users should move their schema files.