Hw to make sure AI Agents neer forget the Work?

This GitHub repository addresses a core challenge in AI-assisted software development: LLMs (large language models) have limited context windows and often "forget" earlier parts of a coding project, leading to inconsistent or broken code. The repo provides project templates and workflows designed to maintain project context and state across AI coding sessions, helping tools like Cursor, Copilot, or Claude stay coherent across long development sessions. This is part of a broader ecosystem where developers use "agentic" AI tools that need structured memory/context management to be practically useful beyond simple code snippets.