MemoryMesh is a local knowledge graph server designed to help AI models maintain structured memory across conversations, with a focus on text-based RPGs and interactive storytelling. It automatically generates dynamic tools from your schemas, enabling the AI to add, update, and link nodes and edges in a graph. The server includes a SchemaManager tool for creating and editing schemas via a visual interface, and ships with pre-built RPG-focused schemas to accelerate setup. To use MemoryMesh, start the server and interact with the generated tools (e.g., add_npc, update_npc, add_location) through MCP-compliant clients. The server emits events for memory changes, which you can monitor to gain insights into how the knowledge graph evolves during gameplay or conversations. Be sure to place your schema files in the expected dist/data/schemas directory so MemoryMesh can discover and generate the corresponding tools on startup.

Prerequisites:

• Node.js (18+ recommended) and npm
• Git

Install and run locally:

1. Clone the repository: git clone https://github.com/CheMiguel23/MemoryMesh.git
2. Navigate to the project directory: cd MemoryMesh
3. Install dependencies: npm install
4. Build the project (if a build step is required by the project): npm run build
5. Start the MemoryMesh MCP server: npm run start

   or directly: node dist/index.js

Notes:

• If you update schemas or dependencies, re-run the build and restart the server.
• Ensure your schemas are located in dist/data/schemas after build, as MemoryMesh auto-detects them on startup.

Tips and considerations:

• The MCP SDK was updated to v1.25.2 (as of v0.3.0) to align with the latest MCP specification. After updating, run npm install to fetch updated dependencies.
• The default schema location changed in v0.2.7 to dist/data/schemas; migrate any older schema files to this path.
• MemoryMesh auto-generates tools from schemas (e.g., add_npc, update_npc, delete_npc). You can customize tool descriptions and constraints through the schema’s description, properties, and relationships.
• The SchemaManager tool provides a visual interface for schema creation and editing and helps avoid writing JSON manually.
• For best results with AI memory guidance, leverage metadata, enumerated types, and explicit relationship edges (e.g., located_in, owns) within your schemas.
• When deploying with clients (Claude Desktop, ChatGPT, Cursor, Gemini, VS Code), ensure the MCP client and server versions are compatible with the 2025-11-25 MCP specification.
• If you encounter path or schema loading issues, verify that dist/data/schemas contains valid .schema.json files and that the server has read access to the dist directory.