Cursor cut agent token cost 7% by deleting two-thirds of its system prompt.

Their September 23 post reports a 66% smaller system prompt with no quality loss, achieved mostly by removing DO NOT rules that newer models no longer need, validated with A/B tests on production traffic rather than evals. Keeping only read, search, edit and shell static and loading the rest on demand cut static tool-description tokens 60%. Explicit cache breakpoints separating stable layers from the growing conversation cut cold cache misses 20%. Numbering every tenth line in file reads instead of every line saved 1.6% of cache-read tokens. All four port to a home-built harness, and the line-numbering one takes ten minutes.