dsh-file-finder:在 DSH 内按名与内容查找并预览、编辑工作区文件

dsh-file-finder 是 chzisnull 维护的 DSH 客户端插件,在 Harness 内按文件名或内容搜索工作区,支持目录浏览、原生右侧栏预览、自带查看器与编辑器,快捷键习惯对齐 VS Code(Ctrl+P / Ctrl+Shift+F)。读写走 DSH 的 ctx.fs,设计为纯加法不占用核心文件界面。npm 安装:dsh plugin add dsh-file-finder。可选桥接 DS Bot 模式切换。MIT 许可,当前包版本 0.11.0。

Read More
dsh-find-file: A File Search Tool Immune to DSH Permissions

dsh-find-file is a DeepSeek Harness (DSH) plugin designed to resolve issues on Windows where the built-in glob tool stops searches and loses results due to permission errors. It registers a pure Node.js file-search tool as the `find_file` agent, supports wildcard-based file name searching, skips inaccessible directories, and returns a report containing matched results, error statistics, and search completeness status. Installation requires creating a directory junction and updating configuration files. It is mainly intended for Windows ACL scenarios and requires Node.js version 20 or later. It uses the `complete` field to explicitly distinguish a “completed search” from an “interrupted search,” preventing permission errors from being misinterpreted as “file not found.”

Read More
dsh-frecency: A Persistent Index and Frecency-Based File Search Plugin for DeepSeek Harness

`dsh-frecency` is a plugin for DeepSeek Harness that addresses performance bottlenecks caused by frequent process spawning in the built-in `grep` and `glob` tools. By using a resident in-memory index and a frecency-based ranking mechanism, it significantly improves file search efficiency. It overrides the original tool names and schemas, allowing models to use it seamlessly without modifying prompts. The plugin supports semantically equivalent `glob`, search results annotated with Git status, and graceful degradation. Installation requires Node.js >= 22 and is compatible with DSH versions 0.1.2-alpha.3 through 0.2.1-alpha.1.

Read More
DeepSeek Harness (DSH) Plugin: zhourenke/dsh-tool-everything

This plugin provides the DSH model with a file search tool based on the Void Tools Everything engine, addressing the model's low search efficiency and high syntax barrier. Usage requires a Windows environment, an installed Everything instance, and a configured `es.exe` environment variable. Core features include millisecond-level whole-system search, full-text retrieval with metadata retrieval, and full support for the Everything search syntax. Typical usages include searching files by size, path, or content (with path restrictions). Caveats include the inability to detect hard links, lack of support for regular expression grouping, and limitations with paths containing spaces. This tool greatly lowers the barrier for the model to call and access the local file system.

Read More
dsh-sidebar-enhancement-search: Built-in Explorer Search and File Type Badges

The DeepSeek Harness (DSH) plugin, DSH Sidebar Enhancement Search, aims to improve the efficiency of the built-in file manager. It addresses the pain points of file searching and visual differentiation by adding an embedded filter box and colored file-type badges to the sidebar. It uses a subsequence matching algorithm, supports fuzzy search and hover-based folder location, and includes a self-healing mechanism. The author notes a code dependency risk: it requires a specific version of `dsh-better-sidebar`, has not undergone large-scale testing, and the service must be restarted after installation.

Read More
dsh-tool-everything: Integrate Everything full-disk file search for DSH Agent

`dsh-tool-everything` is an Everything search plugin for DSH (DeepSeek Harness) that registers the `everything_search` tool for an Agent. It directly invokes the local Everything NTFS index on Windows to perform millisecond-level full-disk searches by filename, without needing to enable an HTTP server or es.exe. The plugin loads the bundled Everything64.dll (taken from Everything SDK 1.5) via the koffi FFI, communicates with Everything through an IPC channel, and returns structured results (path, size, is a folder). It supports the full Everything search syntax, such as wildcards.

Read More