Show HN: Keep large tool output out of LLM context: 3x accuracy 95% fewer tokens

by loumaciel | View on Hacker News