How gondchawarswaraj45/moviemaxx works
Moviemaxx is a conversational Bollywood movie recommender. A user submits a query through the web UI; the API delegates processing to a coordinator, which audits safety, retrieves and ranks movies using graph and vector signals, then adds streaming and social-watch guidance for the response. Catalog ingestion, recommendation history, and the graph explorer are supporting capabilities. Agent wiring and the Graph RAG scoring path are visible in sampled source; some API routes and agent internals were not sampled, so uncertain links are omitted.
Main entry points: app.py FastAPI · coordinator_agent.py Coordinator Agent · movie_graph.py Movie Graph Ops · graph_rag.py Graph RAG
Components 20
User Experience
- Web UI
static/app.js - FastAPI
app.py - Graph Explorer
static/graph.html
Agent Orchestration
- Coordinator Agent
agents/coordinator_agent.py - Guardrail Agent
agents/guardrail_agent.py - P2 Guardrails
guardrails/p2_guardrails.py - Recommender Agent
agents/recommender_agent.py - Streaming Agent
agents/streaming_agent.py - Watch Planner
agents/watch_planner_agent.py
Recommendation
- Graph RAG
rag/graph_rag.py - Relative Chunker
rag/chunker.py - Cohere Embedder
rag/cohere_embedder.py
Catalog and Data
- Movie Graph Ops
movie_graph.py - Movie Catalog
expanded_dataset.py - Catalog Metadata
metadata/catalog_metadata.py - Agent Skills
metadata/agent_skills.py - Neo4j MCP Client
mcp/neo4j_mcp_client.py
Other components
- Viewer
- Cohere API
- Neo4j
Connections 29
- Viewer → Web UI: submits query
- Web UI → FastAPI: calls API
- FastAPI → Coordinator Agent: processes query
- Coordinator Agent → Guardrail Agent: audits prompt
- Guardrail Agent → P2 Guardrails: evaluates prompt
- Coordinator Agent → Recommender Agent: requests candidates
- Recommender Agent → Graph RAG: gets recommendations
- Graph RAG → Movie Graph Ops: searches graph
- Movie Graph Ops → Neo4j: queries catalog
- Graph RAG → Relative Chunker: chunks movie data
- Graph RAG → Cohere Embedder: embeds query
- Cohere Embedder → Cohere API: requests embeddings
- Coordinator Agent → Streaming Agent: gets platform guidance
- Coordinator Agent → Watch Planner: plans watch context
- Coordinator Agent → FastAPI: returns response
- FastAPI → Web UI: serves response
- Viewer → FastAPI: requests catalog sync
- FastAPI → Movie Graph Ops: starts import
- Movie Graph Ops → Movie Catalog: loads records
- Movie Graph Ops → Neo4j: writes catalog
- Viewer → Graph Explorer: opens explorer
- Graph Explorer → FastAPI: requests graph data
- FastAPI → Neo4j: reads graph data
- FastAPI → Catalog Metadata: serves metadata
- FastAPI → Agent Skills: lists skill metadata
- Agent Skills → P2 Guardrails: runs safety skill
- Agent Skills → Movie Graph Ops: runs graph search
- Agent Skills → Graph RAG: runs vector search
- Neo4j MCP Client → Neo4j: queries or writes