章节结构（全文标题提取）： 1 Introduction 2 Forgetting Requires Remembering Why 3 Prompts Ratchet: Growth, Rewrite, Regrowth 4 Comments Halt the Ratchet and Buy Back Instruction-Following 5 Related Work 6 Discussion Limitations Use of Large Language Models Ethics Statement 3.1 Agentic prompts grow unbounded until a wholesale rewrite · 3.2 Most disappearances are not deletions · 3.3 Growth immediately resumes after rewrite · 3.4 Deletion hazard falsifies rival mechanisms · 4.1 Inverting IFEval makes the minimum cover observable · 4.2 Comments encoding latent reasoning settle the prompt at its cover · 4.3 In real prompts, noisy instructions cost correctness and comments buy it back · A.1 Corpus construction
📊 结果
"If English is the new code, why don't we have comments yet?" —— 给 agentic prompt 写注释，是抗膨胀的关键杠杆。
Agentic coding READMEs like CLAUDE.md grow without bound in real repositories, stopping only when the repository retires or someone rewrites the file wholesale. We trace this to imperfect recall: appending an instruction is always cheap, but once an instruction's rationale is gone, deleting it without risking a correctness regression costs O(2^|D|) in a prompt of |D| instructions. We name the resulting divergence catastrophic remembering, the inverse of catastrophic forgetting around which continual learning is organized. First, we characterize this phenomenon across 247,694 instruction lifetimes in 1,867 repositories: agentic prompts grow without bound, more than tripling over their lifetime (+226%), gaining +4.9 net instructions every commit; further, the older an instruction gets, the less likely it is to be deleted (log-hazard -0.032/commit). Then, we show that prompt comments can halt the growth: inverting IFEval yields verifiable worlds whose optimal prompts are known, and there comments encoding latent reasoning remove 99.3% of excess instructions (+211.3% to +1.4%). Finally, applying the same inversion to WildIFEval, we show that prompt comments can improve real-world agentic instruction-following by up to 23.1%. If English is the new code, why don't we have comments yet?