OpenGIN-bot project

[sehansi-9]

This pr introduces the OpenGIN AI agent chat bot

[gemini-code-assist]

Code Review

This pull request introduces the OpenGIN Bot, a full-stack application featuring a Next.js frontend and a FastAPI backend powered by LangGraph for querying temporal graph databases. The implementation includes sophisticated context management strategies like topic shift detection and fact distillation. Feedback highlights several critical issues: a session leakage risk caused by global variables in the Next.js route handler, potential runtime errors in the graph API client when handling 500 responses, and security concerns regarding insecure CORS policies and sensitive information leakage in error responses. Additionally, corrections are needed for hardcoded environment URLs, invalid dependency versions in package.json, and incorrect LLM model identifiers.

[gemini-code-assist]

Next.js version 16.0.10 is invalid as the current major version is 15. This typo will prevent successful dependency installation.

Suggested change

	"next": "16.0.10",
	"next": "^15.1.0",

[gemini-code-assist]

The GraphAPIClient should be properly closed to prevent connection leaks. While a FastAPI lifespan handler is one way to manage this, in low-throughput contexts it is also acceptable to instantiate the client within a context manager for each request. This ensures safe, automatic cleanup with minimal overhead.

References

1. In low-throughput contexts where HTTP requests are made sequentially at a human pace, it is acceptable to instantiate a new httpx.Client for each request using a context manager. The negligible performance overhead is outweighed by the benefit of safe, automatic cleanup.