toolshrink
按内容语义压缩工具结果：13 个 reducer 在 tools/post-execute 阶段按含义裁剪超长输出（保留失败的测试、丢弃 diff 上下文、对 JSON 与 CSV 抽样、lint 问题按规则归类、依赖库的调用栈帧折叠、文件过多的目录只给计数），完整原文落盘保存并留一个定位符可取回。
Content-aware tool-result reducer: 13 reducers cut oversized output by meaning at tools/post-execute (failing tests kept, diff context dropped, JSON and CSV sampled, lint problems grouped by rule, dependency stack frames collapsed, crowded directories counted), with full originals spilled to disk behind a locator.
怎么安装
dsh plugin add github:unclecode/toolshrink git clone https://github.com/unclecode/toolshrink ~/.claude/skills/toolshrink 简介
toolshrink **Cut large agent tool output by what it means, not by where it was cut.** I use Claude Code every day, and I always wanted to intervene in how it manages context. In the early days you could edit the session JSONL directly. Then that door closed. When DeepSeek open-sourced Harness, where everything is a plugin, I looked inside. Tool output there is cut by size: keep the head, keep the tail, drop the middle. I read Codex and pi, and they do the same. None of them look at what the text contains. That fails in a predictable way. Your test suite prints 5,000 passing lines and 3 failure…
推荐参考
信息
- 协议
- MIT
- 语言
- TypeScript
- GitHub 星标
- 10
- 月下载
- –
- 最近更新
- 2026-08-19
- 创建于
- 2026-08-17
基础安全检查
- 检查结果
- curated 收录但无 npm 包/安装命令
- 收录来源
- curated:awesome-dsh-plugin.com, curated:awesome-dsh-plugin/awesome-dsh-plugin
- 主题标签
- ai-agents, claude-code, context-engineering, deepseek-harness, developer-tools, dsh-plugin, llm