debezium/dbz#2156 feat: add reactive LangGraph agent demonstration ex…

[KMohnishM]

…ample
Completes debezium/dbz#2156

Description

This Pull Request implements the second half of GSoC Week 6 targets under tracking issue debezium/dbz#2156. It introduces the reactive_agent example under examples/reactive_agent/, which demonstrates a LangGraph agent reacting dynamically in real-time to change data capture (CDC) updates from the primary PostgreSQL database.

Key Changes

1. Containerized Database Environment:
   • Added docker-compose.yml defining a PostgreSQL service pre-configured for WAL logical replication (wal_level=logical) and initial seeding schema (setup_db.sql).
2. Sync Daemon Ingestion:
   • Implemented stream_sync.py wrapping the embedded Debezium engine to capture logical transactions on catalog rows and synchronize them to Chroma in real-time.
3. Agent & Tool Binding:
   • Implemented agent.py creating a LangChain retriever tool using create_retriever_tool. Supports OpenAI, Ollama, and a fully functional local Mock agent fallback that inspects product quantity levels.
4. Automated End-to-End Demo:
   • Created demo.py performing the full reactive simulation sequence: initial catalog retrieval, executing a SQL update to sell out a product, capturing the WAL log, updating the vector database, and querying the agent again to witness the changed decision.
5. Code Standardizations:
   • Fixed module-resolution duplication in MyPy by establishing standard package boundaries (__init__.py files) in examples/.
   • Updated tool annotations to reference BaseTool across all codebase examples.

Verification Results

• Ruff Quality Check: Passed (All checks passed!).
• MyPy Type Safety: Passed (Success: no issues found in 45 source files).
• Test Suite: Run and passed 86 tests successfully.