Data Flow Pipeline

The system is currently operated sequentially under the server orchestrator. The typical execution data flow proceeds as follows:

graph TD;
    A[User Submits Repository URL via App] --> B(Orchestrator Clones Repo);
    B --> C(Parser);
    C -- "graph.json (Granular Tree)" --> D(Clustering Engine);
    D -- "clustered_structure.json" --> E(LLM Explainer);
    E -- "clustering_data.json (AI Enhanced)" --> F(React Frontend App);
    F --> G[User Explores Visual Graph];

Input/Output Artifacts

1. graph.json
   • Producer: parser.py
   • Consumer: run_clustering.py
   • Description: A highly detailed collection of raw semantic references. Includes lists of nodes (classes, files, methods) and edges (containment, specific method calls).
2. clustered_structure.json
   • Producer: clustering.py
   • Consumer: explainer.py
   • Description: Represents grouped arrays of source code files (cluster_0 = [file1, file2]) and the aggregated weight of external links calling out between clusters.
3. clustering_data.json
   • Producer: explainer.py
   • Consumer: app/public/
   • Description: The ultimate human-readable end product. Consists of a global system_story, the pipeline array of individual AI-explained cluster definitions (title, role, inputs/outputs, interesting patterns), and a flow array denoting the mathematical connectivity.