Move SIC example to demos repo

README.md

@@ -141,6 +141,15 @@ knowledge graphs seamlessly within Memgraph.
 - **:bulb: Demo: [Graph-Aware Agents with LangGraph and Memgraph AI Toolkit](./integrations/langgraph/memgraph-toolkit-chatbot)**
   - This demo showcases a simple agent built using the LangGraph framework and the [Memgraph AI Toolkit](https://github.com/memgraph/ai-toolkit) to demonstrate how to integrate graph-based tooling into your LLM stack.
 
+**MCP**
+- **:bulb: Demo: [SIC classification agent](./integrations/mcp/sic-agent)**
+  - This demo showcases a FastMCP agent that classifies a free-form business description into the OSHA SIC taxonomy stored in Memgraph.
+  - **:mag_right: Key Features:**
+    - Vector search over `IndustryGroup` nodes with Memgraph's vector index
+    - Context expansion through neighboring `Industry` and `MajorGroup` nodes
+    - Clarifying follow-up questions when the first match is ambiguous
+    - Included scraper and embedding scripts for building the SIC graph data
+
 **LlamaIndex**
 - **:bulb: Demo: [KG creation and retrieval](./integrations/llamaindex/property-graph-index)**
   - This demo demonstrates the use of LlamaIndex with Memgraph to

integrations/mcp/sic-agent/README.md

@@ -0,0 +1,56 @@
+# SIC Classification MCP Agent
+
+This example ports the SIC classification agent from the Memgraph AI Toolkit
+into this repository as a standalone FastMCP server.
+
+The agent:
+- embeds a free-form business description,
+- performs vector search over `IndustryGroup` nodes in Memgraph,
+- inspects nearby `Industry` and `MajorGroup` context,
+- uses MCP sampling to select the best SIC code,
+- asks a clarifying follow-up when the initial match is ambiguous.
+
+## Files
+
+- `sic_classification.py` runs the FastMCP server.
+- `sic-scrapper/main.py` scrapes the OSHA SIC manual and generates import
+  Cypher.
+- `sic-scrapper/embeddings.py` generates embeddings for `IndustryGroup` nodes.
+- `sic-scrapper/output/sic_vector_index.cypherl` creates the vector index used
+  by the server.
+
+## Quick Start
+
+1. Generate SIC data and embedding updates:
+
+```bash
+cd sic-scrapper
+uv sync
+uv run main.py
+uv run embeddings.py --from-json output/sic_data.json
+```
+
+2. Load the generated data into Memgraph:
+
+```bash
+mgconsole < output/sic_import.cypherl
+mgconsole < output/sic_embeddings.cypherl
+mgconsole < output/sic_vector_index.cypherl
+```
+
+3. Run the MCP server:
+
+```bash
+cd ..
+uv sync
+uv run python sic_classification.py
+```
+
+## Environment Variables
+
+- `MEMGRAPH_URL` defaults to `bolt://localhost:7687`
+- `MEMGRAPH_USER` defaults to an empty string
+- `MEMGRAPH_PASSWORD` defaults to an empty string
+- `MEMGRAPH_DATABASE` defaults to `memgraph`
+- `SIC_VECTOR_INDEX` defaults to `sic_industry_group_embedding`
+- `SIC_EMBEDDING_MODEL` defaults to `all-MiniLM-L6-v2`

integrations/mcp/sic-agent/init.bash

@@ -0,0 +1,7 @@
+#!/bin/bash
+
+set -euo pipefail
+
+curl -LsSf https://astral.sh/uv/install.sh | sh
+uv sync
+uv run python sic_classification.py

integrations/mcp/sic-agent/pyproject.toml

@@ -0,0 +1,11 @@
+[project]
+name = "sic-classification"
+version = "0.1.0"
+description = "SIC classification MCP server using Memgraph vector search"
+readme = "README.md"
+requires-python = ">=3.10"
+dependencies = [
+    "fastmcp>=0.1.0",
+    "neo4j>=5.28.1",
+    "sentence-transformers>=2.2.0",
+]

integrations/mcp/sic-agent/sic-scrapper/README.md

@@ -0,0 +1,47 @@
+# SIC Scraper
+
+Scrapes the SIC (Standard Industrial Classification) hierarchy from OSHA and
+generates Cypher queries for Memgraph import.
+
+## SIC Hierarchy
+
+```text
+Division (A-J) -> Major Group (2-digit) -> Industry Group (3-digit) -> Industry (4-digit)
+```
+
+## Usage
+
+```bash
+uv sync
+
+# Full scrape
+uv run main.py
+
+# Quick scrape (skip industry details)
+uv run main.py --no-industry-details
+```
+
+## Output
+
+- `output/sic_data.json` - Complete hierarchy
+- `output/sic_import.cypherl` - Cypher import queries
+- `output/sic_embeddings.json` - Generated IndustryGroup embeddings
+- `output/sic_embeddings.cypherl` - Embedding update queries
+
+## Import to Memgraph
+
+```bash
+mgconsole < output/sic_import.cypherl
+mgconsole < output/sic_embeddings.cypherl
+mgconsole < output/sic_vector_index.cypherl
+```
+
+## Generate Embeddings
+
+```bash
+uv run embeddings.py --from-json output/sic_data.json
+```
+
+## Data Source
+
+https://www.osha.gov/data/sic-manual