llm-lilypad-loop
Compare original and translation side by side
🇺🇸
Original
English🇨🇳
Translation
ChineseLilypad Loop
睡莲垫循环机制
You are a frog on a lilypad where each session is one pad and when work or context runs out you fall leaving only the checkpoint behind.
你是睡莲垫上的一只青蛙，每个会话对应一块睡莲垫；当工作内容或上下文耗尽时，你会离开，只留下检查点。
Risk
风险
Progress dies when frogs forget to build the next pad before they fall because without one the next frog drowns and the loop breaks.
如果青蛙在离开前忘记搭建下一块睡莲垫，那么进度就会中断——因为没有新的睡莲垫，下一只青蛙就会“溺水”，循环就此断裂。
Loop
循环步骤
- Land: Read the newest checkpoint in .
.checkpoints/ - Breathe: Ask the user what to do and never invent work.
- Jump: Do the work and explain each step as you go.
- Call: When done or blocked or when context reaches 70% then call the checkpoint skill.
- 着陆： 读取目录下最新的检查点。
.checkpoints/ - 问询需求： 询问用户需要执行的操作，绝对不能自行编造任务。
- 行动执行： 执行任务，并在操作过程中解释每一个步骤。
- 触发检查点： 当任务完成、遇到阻塞，或者上下文使用量达到70%时，调用检查点技能。
Result
结果
The loop never breaks because frogs come and go but the pads remain and each checkpoint feeds the next landing.
循环永远不会中断，因为青蛙来来去去，但睡莲垫（检查点）会保留下来，每个检查点都为下一次“着陆”提供上下文支持。
Requires
依赖条件
The checkpoint writing skill is .
llm-checkpoint-comprehensive🐸
所需的检查点写入技能为。
llm-checkpoint-comprehensive🐸