update

Author: endomorphosis

MCP_SERVER.md

@@ -0,0 +1,136 @@
+# IPFS Datasets MCP Server Integration
+
+This document outlines the Model Context Protocol (MCP) server implementation for IPFS Datasets Python.
+
+## Overview
+
+The IPFS Datasets MCP server provides a standardized interface for AI models to interact with IPFS datasets. It implements the Model Context Protocol, allowing AI assistants to perform operations like:
+
+- Loading datasets from various sources
+- Saving datasets to IPFS
+- Converting dataset formats
+- Processing datasets with transformations
+- Querying dataset content
+- Managing IPFS interactions
+- Vector search operations
+- Knowledge graph interactions
+- Audit logging
+- Security operations
+- Provenance tracking
+
+## Architecture
+
+The MCP server implementation consists of:
+
+1. **Core Server**: Implements the MCP protocol using either the `modelcontextprotocol` package or a simplified Flask-based implementation.
+
+2. **Tool Categories**:
+   - `dataset_tools`: Tools for dataset operations
+   - `ipfs_tools`: Tools for IPFS interactions
+   - `vector_tools`: Tools for vector operations and similarity search
+   - `graph_tools`: Tools for knowledge graph operations
+   - `audit_tools`: Tools for audit logging
+   - `security_tools`: Tools for security operations
+   - `provenance_tools`: Tools for tracking provenance
+   - `cli`: Command-line interface tools
+   - `functions`: Function execution tools
+
+3. **Configuration System**: Flexible configuration via YAML files
+
+4. **IPFS Kit Integration**: Built-in integration with `ipfs_kit_py`
+
+## Getting Started
+
+### Installation
+
+The MCP server is included in the IPFS Datasets Python package.
+
+```bash
+pip install ipfs-datasets-py
+```
+
+### Running the Server
+
+You can start the server using:
+
+```bash
+cd /path/to/ipfs_datasets_py
+./ipfs_datasets_py/mcp_server/start_server.sh
+```
+
+Or for the simplified implementation:
+
+```bash
+cd /path/to/ipfs_datasets_py
+./ipfs_datasets_py/mcp_server/start_simple_server.sh
+```
+
+### Demo Script
+
+We provide a demo script that starts the server and tests its functionality:
+
+```bash
+./demo_mcp_server.py
+```
+
+## Available Tools
+
+The server provides tools in the following categories:
+
+1. **Dataset Tools**:
+   - `load_dataset`: Load a dataset from a source
+   - `save_dataset`: Save a dataset to a destination
+   - `convert_dataset_format`: Convert a dataset between formats
+   - `process_dataset`: Apply transformations to a dataset
+
+2. **IPFS Tools**:
+   - `get_from_ipfs`: Get content from IPFS
+   - `pin_to_ipfs`: Pin content to IPFS
+
+3. **Vector Tools**:
+   - `create_vector_index`: Create a vector index from dataset
+   - `search_vector_index`: Search a vector index
+
+4. **Graph Tools**:
+   - `query_knowledge_graph`: Query a knowledge graph
+
+5. **Audit Tools**:
+   - `record_audit_event`: Record an audit event
+   - `generate_audit_report`: Generate an audit report
+
+6. **Security Tools**:
+   - `check_access_permission`: Check access permissions
+
+7. **Provenance Tools**:
+   - `record_provenance`: Record provenance information
+
+8. **CLI Tools**:
+   - `execute_command`: Execute a command
+
+9. **Function Tools**:
+   - `execute_python_snippet`: Execute a Python code snippet
+
+## Integration Tests
+
+To run the integration tests:
+
+```bash
+python test_mcp_integration.py
+```
+
+This will verify:
+- The server component structure
+- Core functionality
+- Tool availability
+- IPFS Kit integration
+
+## Integration with AI Assistants
+
+AI assistants like Claude can interact with datasets through this MCP server implementation, enabling capabilities like:
+
+- Loading and analyzing data from IPFS
+- Processing datasets with specified transformations
+- Performing similarity searches using vector indices
+- Querying knowledge graphs
+- Generating data visualizations
+- Recording audit events and provenance information