review_assistant

Key Context:

User wants to implement Retrieval-Augmented Generation (RAG) and structured prompting techniques with limited resources and free-tier AI models. User is working with free versions of AI models (ChatGPT, Claude) that have: Small context windows No file upload capability No persistent memory across sessions User's hardware setup: Chromebook with limited internal Linux storage Mount and leverage external storage User is handling large files (2+ million tokens) Primary Objective To create a system that allows the user to effectively review and reason over large prompts, code, and text files despite the limitations of free-tier AI models, by leveraging RAG and structured prompting techniques (MCP, ACP, A2A).

Summary:

The user is seeking to build a workflow that allows them to: Store files, code fragments, and instructions on their external USB drive Strategically access and feed this information to limited AI models Use advanced prompting techniques to maintain context and guide the AI through complex reasoning tasks Effectively manage large projects that exceed normal context limitations of free models The challenge is designing an efficient system that works within these technical constraints while still enabling comprehensive code review and analysis capabilities.