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

(github.com) by loumaciel | Mar 5, 2026 | 1 comments on HN