SMMS Semantic Map MCP Server

The SMMS Semantic‑Map MCP Server is a purpose‑built backend that turns rich 3D instance data into an AI‑ready knowledge base. By treating each object in a semantic map as a first‑class entity, the server offers CRUD (create, read, update, delete) operations on object metadata and builds a searchable index that supports Retrieval‑Augmented Generation (RAG). This eliminates the need for custom data pipelines, allowing developers to focus on building higher‑level AI applications instead of wrestling with raw 3D data.

At its core, the server exposes a simple yet powerful set of resources. Each semantic object—whether a chair, a door, or an appliance—is stored with attributes such as type, spatial coordinates, and semantic tags. The database layer automatically synchronizes these records with the underlying 3D scene, ensuring that any change in the environment is reflected in real time. Developers can then query objects by name, location, or relationship, enabling context‑aware reasoning in conversational agents.

One of the standout capabilities is the built‑in RAG engine. By indexing object attributes and relationships, the server can retrieve the most relevant pieces of information when an AI assistant receives a query. For example, if a user asks “What’s the nearest charging station?” the server quickly returns the relevant object data, which Claude can then weave into a natural‑language answer. This tight coupling of spatial knowledge and language generation greatly enhances the assistant’s usefulness in navigation, maintenance, or interior design scenarios.

Beyond simple retrieval, SMMS introduces an object‑level cognitive map generator. By analyzing task requirements—such as “assemble a piece of furniture”—the server constructs a cognitive topology that outlines the necessary steps and object interactions. This task‑driven map can be fed directly into planning modules or visualized for human operators, bridging the gap between raw data and actionable plans. The result is a system that not only knows where objects are but also how they fit into higher‑level workflows.

For developers building AI assistants, integrating SMMS is straightforward. The MCP interface exposes standard resource endpoints for object manipulation and RAG queries, while the cognitive map generator offers a dedicated prompt that can be invoked via an AI tool call. This seamless integration means that conversational agents can request real‑time spatial insights, receive structured task plans, and even update the underlying 3D model—all through declarative prompts. In environments such as smart factories, autonomous robotics, or augmented reality experiences, the SMMS Semantic‑Map MCP Server provides a scalable, AI‑centric foundation that turns complex 3D scenes into actionable knowledge.