请教下ClaudeCode在处理复杂任务时怎么实现自动压缩上下文，避免出现: API Error: 400 Your input exceeds the context window of this model. Please adjust your input and try again.
❯ /compact
⎿ Error: Error during compaction: API Error: 400 Your input exceeds the context window of
this model. Please adjust your input and try again.
❯ /context
⎿ Context Usage
⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛀ ⛀ gpt-5.5
⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ 270.8k/200k tokens (135%)
⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁
⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ Estimated usage by category
⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ System prompt: 1.5k tokens (0.7%)
⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ System tools: 14.6k tokens (7.3%)
⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ Memory files: 1k tokens (0.5%)
⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ Skills: 1.3k tokens (0.7%)
⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ Messages: 269k tokens (134.5%)
⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁ ⛁
另外，gpt5.5不是1M的上下文吗？怎么在claudecode中设置上下文长度呀？
8 个帖子 - 4 位参与者
推荐阅读
IT早报:0602：5 月汽车销量汇总：比亚迪、上汽、吉利前三；曝豆包在 6 月下旬上线付费内容；英伟达进军 PC 笔记本处理器；软银超丰田成日本市值最高公司..
推荐阅读
黑客诱骗 Meta AI 客服，盗取多名 Instagram 用户账号
来源: LinuxDo 最新话题查看原文