dsh-encoding: Encoding Error Correction and Garbled Text Recovery for DeepSeek Harness
DeepSeek Harness (DSH) produces garbled Chinese text on Windows due to forced decoding. The dsh-encoding plugin resolves this issue by providing standalone `encoding_run` (execute and automatically decode) and `encoding_decode` (repair garbled text) tools. It is zero-dependency, pure ESM, and requires no build, while correctly handling UTF-8, UTF-16LE, and GBK encodings. It does not replace built-in tools but operates as a supplement; users should note its detection limitations and data size limits.
Read Moredsh-file-download: DeepSeek Harness File and Folder Download Plugin
`dsh-file-download` is a community-maintained standalone plugin designed to address the cumbersome file download process in DeepSeek Harness. It adds convenient download buttons to the Files tree and document preview, supporting the download of individual files or packaging entire folders, including nested structures, into a ZIP archive. Installation is straightforward: download the Tgz package and run the dsh plugin command. The plugin is powerful and secure, supporting streaming for large files, cancellation, and permission inheritance, though ZIP generation is subject to limits on size, entry count, and nesting depth.
Read Moredsh-9router-web-search: 9router Backend Web Search and Retrieval
`dsh-9router-web-search` is a plugin for DeepSeek Harness (DSH) that adapts it to the 9router gateway, designed to address DSH's default dependency on Anthropic's search tools. The plugin directly mounts DSH's native `web_search` and `web_fetch` tools onto 9router-compatible endpoints, enabling live web search and content fetching. It supports zero-config Wiring, credential awareness, and a pure ESM implementation. After installation, configure the 9router Base URL in the settings panel. The plugin requires DSH version `>= 0.1.7-rc.1` and expects 9router to have appropriate security protections in place.
Read Moredsh-element-context: Selecting UI elements in a conversation and injecting them into the model context
The DeepSeek Harness plugin `dsh-element-context` addresses the challenge of models understanding the location and styling of page controls. By selecting elements with the mouse or manually entering CSS selectors, the plugin injects structured information about the target element—such as its selector, box model, and computed styles—into the model context, eliminating the need for screenshots or verbal descriptions. It supports continuous selection, chip management, and sandbox-friendly operations, making it suitable for web automation and debugging scenarios. Installation is straightforward and it does not depend on additional services, but users should be mindful of token consumption per conversation turn and the 32-element limit.
Read MoreDeepSeek Harness Plugin: Retry, Edit Prompt, and Continue Actions for Chat Turns
The `locietta/dsh-rewind-retry` plugin adds Retry, Edit prompt, and Continue actions to DeepSeek Harness (DSH). At its core, it works by forking a conversation: Retry and Edit fork at turn boundaries to resubmit or modify the prompt, while Continue simply continues in the current conversation. The plugin can be installed directly from GitHub; it supports only manually initiated turns, and Retry is limited to text, requiring attachments to be added manually.
Read MoreDeepSeek Harness Plugin: localsummer/dsh-group-chat
The `localsummer/dsh-group-chat` plugin provides DSH with a multi-model, role-based group chat panel, supporting different roles bound to different models and shared context. It allows creating groups and sessions, managing role configurations (such as persona, temperature, and deep thinking), and enables multi-turn conversations with streaming output. Users can use Markdown rendering, image input, and file workspace tools. Data is stored in `~/.dsh/storages/group-chat/`, with support for backup and recovery from corruption.
Read MoreDeepSeek Harness Plugin: ln1m/dsh-tool-literature (Literature Search Tool)
DeepSeek Harness developer Ln1m introduced the `dsh-tool-literature` plugin to address the inefficiency of agents directly executing shell commands. The plugin encapsulates the literature search workflow and exposes a structured interface for direct model invocation. Built on the `paper-search.py` script, it integrates OpenAlex citation ranking with an arXiv relevance scoring algorithm. It depends on libraries such as `@deepseek-ai/cordis` to support automated literature retrieval. Notably, the repository has been archived and merged into `dsh-tool-suite` and is retained for historical reference only.
Read Moredsh-local-file-search: Searching Local Files in a Dialog
This plugin aims to provide a quick local file lookup solution for DeepSeek Harness that does not rely on indexing and does not occupy the file pane. It performs a full-machine scan through an entry in a dialog, with results used only for temporary lookups. The plugin is currently archived, and its content has been merged into dsh-tool-suite. It is recommended to review the source code and license before installation.
Read MoreDeepSeek Harness Plugin: dsh-lt-tasks (Multi-Window Long-Term Task Management)
dsh-lt-tasks is an archived plugin for DeepSeek Harness (DSH) that addresses the loss of context when long-term task sessions end. It achieves multi-window task management and state continuity through a combination of “persistent folders” and “minimal handoff documents.” Built on ES Modules and using React to inject UI components, the plugin is currently hosted in the GitHub repository [dsh-side-suite](https://github.com/Ln1m/dsh-side-suite). Since it is archived and has no formal affiliation with the official project, users should be aware of potential permission risks and follow community updates.
Read MoreDeepSeek Harness Session Archive Button Plugin
The `dsh-archive-button` plugin is an automated session cleanup tool for DeepSeek Harness (DSH), designed to address the issue of historical sessions taking up storage space. Located in the sidebar, it provides a "dry run" prescan for sessions idle for more than 3 days and a secondary confirmation before packaging, creating a zip backup and deleting the original files to ensure data safety. The project has now been archived in the `dsh-chrome-suite` repository and is retained for historical reference only.
Read Moredsh-aurora-glass: An Aurora Glass Background Beautification Plugin for DeepSeek Harness
`dsh-aurora-glass` is a lightweight independently developed plugin designed specifically for DeepSeek Harness (DSH), enhancing the visual hierarchy of the Web UI through aurora backgrounds and glassmorphism effects. It avoids performance-intensive blur pipelines and supports preset color palettes, custom local and remote images, theme-following, and a standalone settings panel. By dynamically injecting styles via Cordis, it maintains smooth interface performance, is compatible with DSH 0.2.x versions, and fully restores the default appearance after uninstallation.
Read MoreDeepSeek Harness Plugin Collection: dsh-bundles
dsh-bundles is a plugin management tool for DeepSeek Harness, designed to address the cumbersome manual installation of community plugins. It provides standardized toolkits categorized by persona, with support for one-click installation, preview, safe uninstallation, and status inspection. The tool includes six curated plugin packs, features a secure rollback mechanism, and offers advanced capabilities such as adopting legacy installations and testing from source. Users need to install it via the command line and enable it in the settings. When using it, attention should be paid to permission security and configuration file paths. The project lowers the barrier to entry through standardized workflows, with its homepage at dshbundles.com.
Read MoreDeepSeek Harness Plugin: dsh-jev
dsh-jev is a plugin for DeepSeek Harness (DSH) designed to address the issue of coding agents consuming expensive inference tokens when handling mechanical decisions. By using TypeSafe Jev, it offloads tasks such as classification, scoring, and routing to lightweight models, allowing the primary model to focus on code generation. The plugin provides core features including typed judgments, skill and tool discovery, administrative commands, cost tracking, calibration toolkits, and CI/CD validation. After installation and configuration, users can issue slash commands or API calls to achieve low-cost structured decision-making and logic authoring.
Read MoreDeepSeek Harness Plugin: Integrating JuMeng AI for Image and Video Generation
The DeepSeek Harness plugin `dsh-jumeng-image-video-tool` is a tool integrated with Jumeng AI, designed to streamline the process of generating images or videos within conversations. By configuring an API key, users can directly invoke features such as text-to-image, image-to-image, text-to-video, and image-to-video generation. The system automatically saves the results to disk and returns an accessible URL. The plugin also supports uploading local assets and provides interfaces such as model list queries. Users should note the file size limits (10MB for images and 100MB for videos) and URL expiration, which is approximately one hour.
Read MoreDeepSeek Harness (DSH) Plugin: Windows 11 / WinUI 3 Skin (`dsh-winui-skin`)
`dsh-winui-skin` is a Windows 11 / WinUI 3 skin plugin for DeepSeek Harness (DSH). Built on an "everything is a plugin" philosophy, it reads the system accent color from the registry on the Host side, passes it to the Client side via API, overrides the existing design, and injects 133 Fluent-style tokens. It supports Light, Dark, and Follow System modes. The plugin relies on system fonts, uses CSS to simulate the Mica material, and automatically restores the original appearance upon uninstallation. To install it, add it to the current or a specified profile via a script or CLI.
Read Moredsh-offpeak: DSH Host Plugin for Automatically Managing Peak-Period Sessions and Requests
dsh-offpeak is a core plugin for DeepSeek Harness, designed to intercept requests during API peak periods to avoid rate-limiting failures. It intercepts, before dispatch, requests targeting the official source or explicitly listed models, and supports turn gating and request-level guards. It provides peak/off-peak policy configuration and session continuous-run exemptions, and manages the interception list, time zone, and peak windows through a Web panel. It is compatible with DSH host versions, supports either hold (suspension) or error throwing, and requires Node.js >= 20.
Read MoreDeepSeek Harness Plugin: Prompt Optimizer
`dsh-plugin-prompt-optimizer` is a zero-dependency prompt optimization plugin designed specifically for DeepSeek Harness, aiming to convert vague drafts into clear, acceptance-ready prompts with one click. It supports input box buttons, slash commands (including offline/diagnostic modes), and model tool calls, and includes a fault-tolerance mechanism that automatically falls back to a local rule engine when the model is unavailable. Installation requires Node.js >= 22 and DSH >= 0.2.0-rc.2, and the plugin is ready to use after restarting the application.
Read Moredsh-client-ui-peak-valley: DSH Web Client Peak/Valley Price Status Display Plugin
The DeepSeek Harness plugin `dsh-client-ui-peak-valley` displays the DeepSeek API pricing status in real time at the UI layer. It automatically checks Beijing time every 30 seconds and uses colors (green/orange) to distinguish off-peak from peak pricing, helping developers optimize costs. Installation requires the `dshpm` command and a process restart. It only supports sessions whose model ID contains `deepseek`, and is compatible with light and dark themes.
Read MoreDSH Hermes Memory: A Lightweight Hierarchical Memory Plugin for DeepSeek Harness
DSH Hermes Memory is a lightweight long-term memory plugin for DeepSeek Harness (DSH), designed to address context history bloat and forgetting issues. It manages agent context without requiring a separate database by using hierarchical storage (such as user preferences and project facts) and automatic refinement and compression mechanisms. It supports privacy encryption, tool-based query management, and compatibility with the DSH environment, helping developers quickly build agents with long-term memory capabilities.
Read Moredsh-task-todo: DeepSeek Harness To-Do Task Management Plugin
The `dsh-task-todo` plugin integrates task management into DeepSeek Harness. It supports multiple views (list, kanban, calendar, Gantt chart) and quick-add syntax, and provides a draggable floating-window mode. Agents can read and write tasks through `task_*` tools and the `/todo` command, with support for subtasks, recurring tasks, priorities, and checklist management. Data is globally synchronized and requires Node >= 22.
Read MoreDSH Proxy Routes
`dsh-proxy-routes` is a network proxy routing plugin for DeepSeek Harness (DSH), designed to address the pain point that multiple accounts under the same provider require different proxy policies. The plugin intercepts API requests at the Fetch layer, identifies the account based on the provided key, and enables precise account-level routing. Core features include a named proxy pool supporting SOCKS5/HTTP/HTTPS protocols, graphical configuration with hot reloading, direct-connection mode control, and proxy support for built-in online tools. Configuration is based on existing LLM accounts in DSH; users can specify that particular accounts use a proxy or direct connection, while unconfigured accounts default to direct connection. The plugin does not contain API keys and only matches them in memory, ensuring security.
Read Moredsh-peak-hours: DeepSeek Harness Execution Window Toggle Plugin
The DeepSeek Harness plugin `dsh-peak-hours` helps users reduce model invocation costs by automatically running or pausing tasks during off-peak hours to optimize resource usage. It supports two modes, "Always Run" and "Run During Off-Peak Hours Only," and determines peak and off-peak windows for Monday through Friday based on the Beijing time zone (UTC+8), with support for statutory holidays. Users can switch modes, query status, and simulate the clock through UI controls or the provided HTTP API. Implemented purely in JavaScript, the plugin should be installed carefully to avoid mixing installation methods. It controls task execution via an interception mechanism, making it suitable for scenarios that require running tasks only within specific time windows.
Read Moredsh-intent-suggest: DeepSeek Harness Plugin for Real-Time Task Candidate Generation
The `dsh-intent-suggest` plugin addresses the issue of users waiting for model responses by generating three candidate tasks in real time above the input box. It supports three categories—coding, research, and custom—and infers intent from the context of the most recent six conversation turns. Clicking a candidate fills it back into the input box without automatically sending it. Users must install the plugin and restart the dsh process, and it also supports custom profiles.
Read Moredsh-git: Integrating the Local Git Workbench and GitHub Tab into the DeepSeek Harness Right Sidebar
`dsh-git` is a plugin for DeepSeek Harness (DSH) that provides local Git operations and remote GitHub management in the right sidebar, reducing window switching for developers. It integrates a local workbench, multi-repository management, a remote GitHub tab, and LLM-powered commit message generation. By combining host-side logic with browser-side UI, it ensures secure token storage and enforces secure push methods. Developers can complete the entire workflow—from code changes and commits to remote repository management—directly within the DSH environment.
Read Moredsh-memory: A Cross-Session Long-Term Memory Plugin for DSH
`dsh-memory` is a local memory plugin for DeepSeek Harness (DSH) designed to address models' lack of cross-session memory. Built on local file-system storage and operating without LLM calls, it uses pure string matching and hash-based deduplication to enable zero-cost long-term memory persistence. Core features include multi-dimensional retrieval, conflict detection with a human adjudication mechanism, and audit logs. It injects memory into context through an L1, L2, and L3 three-tier structure, allowing the model to read and write memory on demand via tool calls. The plugin also provides a settings panel to manage memory, view usage, and resolve conflicts, making it well suited for workflows that require maintaining project context and preferences.
Read More