Dsr MCP Server

DSR MCP Server

An MCP (Model Context Protocol) server that provides a comprehensive set of tools to query, manipulate, and analyze Deep State Representation (DSR) graphs. This server enables AI assistants to interact with robot perception and scene understanding data through standardized MCP tools with real-time graph operations and intelligent node relationship management.

Tools

Resources

Resources provide read-only, efficient access to DSR graph data. They are ideal for querying information without modifying the graph state.

Environment Variables

Installation

Dependencies

• [fastmcp](https://github.com/jlowin/fastmcp): MCP server framework (>= 2.2.7)
• [Cortex](https://github.com/grupo-avispa/cortex): DSR library for graph operations
• Python: 3.12+ required

Note: You must build Cortex inside the virtual environment to ensure compatibility.

cd dsr_mcp_server

source .venv/bin/activate cd ${CORTEX_DIR} && make -p build && cd build cmake .. && make -j$(nproc) && sudo make install

Install with uv (recommended)

Clone the repository and install with uv:

git clone https://github.com/grupo-avispa/dsr_mcp_server.git

cd dsr_mcp_server uv sync

Or install directly from the repository:

uv add git+https://github.com/grupo-avispa/dsr_mcp_server.git

Install with pip

Install the package in mode:

git clone https://github.com/grupo-avispa/dsr_mcp_server.git

cd dsr_mcp_server python3 -m pip install .

Or install directly from the repository:

python3 -m pip install git+https://github.com/grupo-avispa/dsr_mcp_server.git

Usage

Running with uv

uv run dsr_mcp_server

Running with pip installation

python3 -m dsr_mcp_server

The server will start and attempt to initialize the DSR connection automatically using the configured parameters.

Configuration example for Claude Desktop/Cursor/VSCode

Using uv (recommended)

Add this configuration to your application's settings (mcp.json):

{

"dsr mcp server": { "type": "stdio", "command": "uv", "args": [ "run", "--directory", "/path/to/dsr_mcp_server", "dsr_mcp_server" ], "env": { "DSR_AGENT_ID": "42", "DSR_NAME": "your_dsr_graph_name" } } }

Using pip installation

{

"dsr mcp server": { "type": "stdio", "command": "python3", "args": [ "-m", "dsr_mcp_server" ], "env": { "DSR_AGENT_ID": "42", "DSR_NAME": "your_dsr_graph_name" } } }

HTTP Server Mode

For HTTP transport integration:

{

"servers": { "dsr_mcp_server": { "type": "http", "url": "http://localhost:3000/mcp" } } }

Technical Notes

• Connection to DSR is performed with automatic initialization and connection monitoring.
• Node attributes are filtered to exclude internal rendering properties (color, depth, height, level, etc.) for cleaner output.
• Edge relationships support various types including spatial (near), ownership (has), identity (is), and association (is_with) semantics.
• Graph operations maintain consistency through the DSR library's built-in validation mechanisms.
• All tools return standardized JSON responses with success/error status and detailed information.
• Interactive file selection for graph export operations through MCP elicit mechanism.
• Resources provide read-only, idempotent access to DSR graph data with efficient caching and lower overhead compared to tools. Use resources for querying data and tools for modifications.

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes following the coding conventions
4. Add tests if applicable
5. Submit a pull request

Support

For issues, questions, or contributions, please refer to the project's issue tracker or contact the development team.