README.md

metadata

title: Code Knowledge Graph Explorer — 🤗 Transformers Library
emoji: 🔍
colorFrom: blue
colorTo: purple
sdk: docker
app_port: 7860
pinned: false
tags:
  - building-mcp-track-enterprise

Knowledge Graph MCP Explorer

This is a Gradio-based interactive tool for exploring code repository knowledge graphs. It provides a web interface to search, navigate, and analyze code relationships using the Model Context Protocol (MCP).

Features

• Search Nodes: Search for code entities, functions, classes, and more using semantic search
• Graph Navigation: Explore relationships between code elements
• Entity Tracking: View declared and called entities within code chunks
• Path Finding: Find paths between different nodes in the knowledge graph
• Subgraph Extraction: Extract and visualize subgraphs around specific nodes
• File Structure: View the hierarchical structure of the repository

Usage

The application loads a pre-built knowledge graph from the HuggingFace Transformers repository. You can:

1. Search: Use the search tab to find relevant code snippets and entities
2. Explore: Navigate through the graph using node IDs
3. Analyze: Get statistics about the code structure and relationships

Technical Details

• Built with Gradio for the web interface
• Uses LanceDB for efficient code indexing and search
• Supports hybrid search (keyword + semantic embeddings)
• Pre-computed embeddings using Salesforce/SFR-Embedding-Code-400M_R model

Data Sources

The application supports loading knowledge graphs from:

1. HuggingFace Hub Dataset (Recommended)

Load directly from a HuggingFace dataset:

python gradio_mcp.py --host 0.0.0.0 --port 7860 --hf-dataset "username/dataset-name"

2. Local JSON File

Use a local JSON file (e.g., multihop_knowledge_graph_with_embeddings.json):

python gradio_mcp.py --host 0.0.0.0 --port 7860 --graph-file data/multihop_knowledge_graph_with_embeddings.json

Creating and Publishing a Dataset

You can save an existing knowledge graph to HuggingFace Hub:

from RepoKnowledgeGraphLib import RepoKnowledgeGraph

# Load from local file
kg = RepoKnowledgeGraph.load("path/to/graph.json")

# Push to HuggingFace Hub (without embeddings to reduce size)
kg.to_hf_dataset("username/my-knowledge-graph", save_embeddings=False, private=False)

# Or with embeddings (larger dataset)
kg.to_hf_dataset("username/my-knowledge-graph-with-embeddings", save_embeddings=True)

Docker Configuration

The default Dockerfile uses a local JSON file. To use HuggingFace datasets instead, modify the CMD line in Dockerfile:

# Using HuggingFace dataset (recommended for smaller Docker image)
CMD ["python", "-u", "gradio_mcp.py", "--host", "0.0.0.0", "--port", "7860", "--hf-dataset", "username/dataset-name"]

# Using local file (requires large data file in image)
CMD ["python", "-u", "gradio_mcp.py", "--host", "0.0.0.0", "--port", "7860", "--graph-file", "/app/data/multihop_knowledge_graph_with_embeddings.json"]

Local Development

To run locally:

docker build -t gradio-mcp-space .
docker run -p 7860:7860 gradio-mcp-space

Or without Docker:

pip install -r requirements.txt
python gradio_mcp.py --host 0.0.0.0 --port 7860 --hf-dataset "username/dataset-name"

Deployment to HuggingFace Spaces

Option 1: Using HuggingFace Dataset (Recommended)

1. First, push your knowledge graph to a HuggingFace dataset
2. Update the Dockerfile CMD to use --hf-dataset
3. Push to the Space repository (no large files needed)

Option 2: Using Local JSON File

1. Create a new Space on HuggingFace with Docker SDK
2. Enable Git LFS in your Space repository
3. Push this directory to the Space repository:

   git lfs install
   git lfs track "data/*.json"
   git add .
   git commit -m "Initial commit"
   git push
   

👥 Team

Team Name: CEPIA Ionis Team

Team Members:

• Laila ELKOUSSY - @lailaelkoussy - Research Engineer, Data Scientist
• Julien PEREZ - @jnm38 -