Engineering Practice: Managing Massive Project Context for LLMs
throwaway0134hdj · reddit · 2026-08-02
A developer asks how to efficiently use LLMs to understand and manage massive project documents. Faced with over 500+ PDFs, Word, and Excel files, the author wants to query the entire corpus at any time and get accurate answers, while also using the context to write code rules and SQL table structures.
Currently, they dump everything into a single folder and use VSCode with the Cline extension to query all documents directly. The community is discussing more efficient and sustainable context organization and retrieval engineering workflows.
More from coding & agent
- Keep Agents Vanilla: Developers Argue Against Over-Configuring AGENTS.md — cnakazawa · 2026-08-03
- DeepSeek Cache Hits Rely on the Model Itself; Agent Tooling Faces Different Challenges — teortaxesTex · 2026-08-03
- GitHub 3.8k stars: Build a modular Agentic RAG system with LangGraph — tom_doerr · 2026-08-03
- From UX to AX: Designing Products for AI Agents — davidhoang · 2026-08-03
- Testing DeepSeek V4 Flash Combined with OpenCode — techNmak · 2026-08-03
- DeepSeek Reportedly Building Dedicated 'DeepSeek Code' Agent to Rival Claude Code — iamaliveix · 2026-08-03