dsh-imggen: A Static Plugin Providing Text-to-Image Output Capabilities for DeepSeek Harness

The `dsh-imggen` plugin for DSH provides text-to-image generation capabilities for Harness. It uses a dual-face architecture, with the free, keyless Pollinations backend supported by default, and also includes support for OpenAI (DALL·E 3). The tool offers a tool card for image generation, along with support for downloads, history gallery viewing, and backend switching. Configurations written to the bundle persist across restarts, and API keys are processed locally only. The project is open source and aims to simplify the image generation workflow.

Read More
DeepSeek Harness Plugin: Real-Time Display of DeepSeek Account Balance with Official Top-Up Support

`dsh-deepseek-balance` is a static plugin designed for DeepSeek Harness (DSH) that addresses the inconvenience of checking account balances. By calling the official API, it displays real-time account balances (including multiple currencies) in the DSH interface and provides a top-up entry point. The plugin supports automatic refresh every 60 seconds, adaptive light and dark themes, and includes security mechanisms by ensuring API keys are used only within local processes and never sent over the network. Users can install it via npm, and it retrieves the API Key in priority order from memory, environment variables, or a credential store. This tool allows developers to manage their accounts without frequently switching between web pages, improving the development experience.

Read More
dsh-usage-skill: Token Usage Analysis and Skill Package Management Plugin

dsh-usage-skill is a client plugin that provides centralized resource and skill management capabilities for the DeepSeek Harness (DSH) Web GUI. This plugin primarily addresses three common requirements: token consumption analysis, provider balance checking, and skill package management. To install it, developers need to download the source code from GitHub and place it in the plugin directory to enable the plugin; currently, the public documentation does not provide specific usage examples. Developers should be aware of potential permission risks and review the source code and license before installation to ensure security.

Read More
dsh-reasoning-effort: Reasoning Effort Level Selector for DeepSeek Harness

`dsh-reasoning-effort` is a DeepSeek Harness (DSH) plugin designed to provide a Codex-like reasoning-effort level selection mechanism. The plugin allows developers to finely control the length and depth of the reasoning chain based on the model's declared specifications, thereby extending model capabilities while balancing performance and cost. It supports native DSH themes and integrates a Codex-style reasoning selector. Since official installation documentation is unavailable, it is recommended to manually obtain and install it from the GitHub source repository, and to verify the source code and license information before use.

Read More
DeepSeek Harness Plugin Introduction: dsh-better-markdown

`dsh-better-markdown` is a Markdown rendering enhancement plugin for the DeepSeek Harness Web client. It leverages the `markstream-react` library to optimize the display of streaming output, supporting code highlighting, tables, and complex formatting. Developers can install the plugin for a better reading experience, but should verify the source code license to ensure security compliance.

Read More
dsh-plugin-lmstudio: DSH Plugin for Integrating Local LM Studio Models

dsh-plugin-lmstudio is a DeepSeek Harness (DSH) plugin designed to simplify the integration of local large models. Through OpenAI-compatible routing, it maps LM Studio's local inference capabilities to DSH's conversation layer, enabling local conversations without consuming cloud API quotas. Core features include using LM Studio models, the default `lm-studio` route, and local port communication. After installation, start the server in LM Studio's Developer mode and load a model, then retrieve the model ID via the terminal and configure it in the DSH configuration file, with hot reloading supported. This plugin is suitable for developers who need to run models locally in an isolated environment or avoid API rate limits.

Read More
DSH Git Memory: Git-Based Long-Term Context for DeepSeek Harness

`dsh-git-memory` is a Git-based memory management plugin for DeepSeek Harness, designed to address the challenge of managing long-term context for agents. The plugin is currently in the compatibility testing phase. Its core features include dynamic context registration, metadata observation, and the ability to view status via human commands and model tools. Installation requires Node.js 24, pnpm 11, and DSH 0.1.0-rc.6, and supports use via a Web profile. As it is still in testing, the current version intentionally does not store memory and lacks persistence, write, or background invocation capabilities. It also imposes strict restrictions on sensitive data and Git operations. Its design provides a foundation for a controlled and auditable long-term memory system.

Read More
Introducing the DeepSeek Harness Plugin: dsh-ecc

The staavanothanh/dsh-ecc plugin serves as an adapter layer for DeepSeek Harness (DSH), aiming to enable reuse of ECC ecosystem capabilities within DSH. It converts ECC agent Markdown resources into DSH-recognizable personas and instructions, and exposes capabilities as skills through a skill provider. The plugin implements the native tools `spawn` and `spawn_fork`, supporting explicit configuration of the model, reasoning strength, and context inheritance. Once installed, developers can directly invoke ECC agent definitions and skills within DSH workflows, enabling flexible sub-agent management.

Read More
staavanothanh/dsh-cbm: DSH Adapter for codebase-memory-mcp

`staavanothanh/dsh-cbm` is a plugin for DeepSeek Harness (DSH) used to integrate `codebase-memory-mcp`. It connects to executable files via stdio, exposing the three tools `search_graph`, `trace_path`, and `get_code_snippet` along with SessionStart hooks. Configuration relies on the `CBM_COMMAND` environment variable, supporting either an absolute path or a system PATH. `CBM_ALLOWED_ROOT` can be used to restrict the indexing scope, and `failOnStartupError` can improve startup fault tolerance. The plugin is based on Node.js >= 22.19.0 and follows

Read More
Introducing the DeepSeek Harness UI Color Tone Plugin

`dsh-ui-tint` is a UI tinting plugin designed specifically for DeepSeek Harness (DSH), supporting both desktop and Web GUI. It enables lightweight interface customization through CSS injection, allowing users to independently adjust the color, opacity, and brightness of the sidebar, chat area, and dialog boxes. The plugin provides a rich set of presets, as well as "color depth" adjustment and automatic adaptation to light and dark themes to ensure font readability. All settings are persistently stored and automatically restored after a restart.

Read More
Introducing the DeepSeek Harness Client Connection Authorization Plugin

`sperictao/dsh-client-connection-authz` is an authentication plugin for DeepSeek Harness (DSH), designed to provide fine-grained authorization for external client connection requests. It intercepts connection requests by injecting `ConnectionRequestAuthorizer`, enforces access control policies, and preserves the original connection behavior. The plugin includes local bypass protection and browser security validation mechanisms, and must be used together with an authentication package that implements the authorization logic. This solution is suitable for scenarios that require strict restrictions on remote access permissions, and is compatible with DSH 0.1.0-rc.x and the dependency range `^0.1.6-alpha.1`.

Read More
dsh-auth-tailscale: A DeepSeek Harness Identity Authentication Plugin Built on Tailscale Serve

`dsh-auth-tailscale` is an adapter plugin for DeepSeek Harness (DSH), providing DSH with identity authentication and fine-grained access control based on Tailscale Serve. It leverages user information injected by Tailscale to enforce user-specific access control, general-use restrictions, and administrative permission limits through allowlists and App Capability. Configuration is set through environment variables, and a fail-closed mechanism is used to ensure security. Deployment requires Tailscale Serve to listen on the loopback address, with DSH bound locally. The plugin has no runtime dependencies and is decoupled from DSH versions.

Read More
dsh-plugin-workspace-skill

The DSH plugin `dsh-plugin-workspace-skill` addresses skill isolation and management for skill reuse across multiple projects. It provides guidance for writing skills and a workspace-level skill isolation mechanism, locating skill directories under `<workspace>/.dsh/skills/` and supporting both directory-package and single-file layouts. By scanning and registering workspaces, workspace skills take precedence over global skills, enabling safe overrides of skills with the same name and standardizing the local Skill development workflow.

Read More
dsh-lan: DSH LAN Exposure and Balance Query Plugin

`dsh-lan` is a DeepSeek Harness (DSH) plugin designed to address the security risks associated with Web runtimes bound only to the local loopback address and API key exposure. It enables LAN access by changing the bind address to `0.0.0.0` and provides a local RPC interface for scripts to securely query balances without uploading the API key. Installation requires editing the configuration file; once started, it can be used within a trusted network.

Read More
dsh-experience-plugin: Model Inference Levels, CLI Simulation, and Everyday-Use Optimization Settings

The DSH plugin `dsh-experience-plugin` is designed to improve the user experience of DeepSeek Harness by addressing issues of scattered configurations and complex management. The plugin provides features such as model reasoning-level injection, CLI behavior simulation (with support for Codex, Claude Code, and others), message timeline preview, collapsible content, and automatic loading of history. All configurations are consolidated in the “Daily Optimization” section of the settings page. Licensed under the MIT License and published to npm, the plugin can be installed via a DSH plugin command or npm. Developers can use it to fine-tune model behavior or manage configurations in a unified way.

Read More
dsh-scenery-background: DeepSeek Harness Daily Rotating Scenery Background Plugin

`dsh-scenery-background` is a pure frontend DSH Web interface beautification plugin maintained by `soslowsnail`. It introduces aesthetic sea-and-mountain backgrounds to make the default monotonous interface more vibrant. The plugin features core functions such as Picture of the Day, a 30-second looping carousel, camera movement effects, and frosted-glass floating panels. It supports light and dark theme adaptation and has no network dependency. After installation, a control button appears in the bottom-right corner, allowing users to switch modes, change images, and adjust clarity. In case of a startup failure, the plugin can be disabled by editing the configuration file.

Read More
DeepSeek Harness Third-Party Plugin Manager

Plugin management in DeepSeek Harness (DSH) currently relies on manually editing configuration files, which is cumbersome. To address this, community developer `sorry123maker` has provided a plugin called “Third-Party Plugin Management.” This plugin enables centralized display and control of third-party plugins in `dsh.profile.bundles` through a Web UI, supporting enabling or disabling plugins via toggles and saving their states as presets for easy switching. Plugin changes take effect immediately through hot reload, without requiring a restart. In addition, new plugins can be added by package name. This tool greatly simplifies the plugin configuration process in DSH and enables visualized management.

Read More
dsh-chatfile: Drag-and-Drop Upload of Any Files in the DSH WebUI

dsh-chatfile is an open-source plugin for DeepSeek Harness WebUI, designed to address the issue of agents being unable to handle regular files. It allows users to upload arbitrary files by dragging and dropping them or clicking a button. Uploaded files are persisted to the session's uploads directory, and their paths are automatically generated in the input box for the agent to read. The plugin features a 50MB per-file limit, a secure random-token download mechanism, and image compatibility. A service restart is required after installation for the changes to take effect.

Read More
dsh-codex-kanban: The AI-Native Kanban Companion for DeepSeek Harness

`dsh-codex-kanban` is a localized Kanban plugin designed specifically for DeepSeek Harness. It stores cards in local JSON files via a tool-based interface, supports defining state transitions from `backlog` to `done`, and integrates Codex subagents to execute code. The plugin emphasizes evidence tracking and security, enabling task management on a local dashboard and forwarding specific tasks to the Harness UI for execution. It is suitable for scenarios that require switching between multiple repositories, with DeepSeek coordinating and Codex handling code-related work. Installation requires Node.js 22.19+ and administrator privileges, while day-to-day use relies on browser interactions and the status menu.

Read More
dsh-bookmarks-dock: A Left-Side URL Bookmark Bar Plugin for DeepSeek Harness

dsh-bookmarks-dock is a plugin that adds left-side bookmark management to DeepSeek Harness (DSH). It addresses the pain point of multitasking with an expandable floating toolbar: when expanded, the page content automatically shifts to the right without covering the main view. The plugin features automatic resolution of genuine favicons, persistent bookmark storage, easy adding and removing, and preset commonly used sites. It is easy to install and takes effect upon restart, making it ideal for developers who need to quickly switch between frequently used development tools in the DSH interface.

Read More
dsh-req-workbench: DeepSeek Harness Requirements Management Workbench

dsh-req-workbench is a plugin for the DeepSeek Harness Web UI designed to convert conversation content into a structured requirements list. Key features include creating and editing requirements and subtasks, importing from conversation turns or pasted Markdown text, and polling every minute to remind users of subtasks that will time out within 24 hours. Installation is performed via pnpm using the official command, after which the Web UI must be restarted to use the plugin. The plugin relies on an ESM-built host; using CommonJS in a Node 24 environment may cause crashes.

Read More
doco-dsh: Native Integration of the Doco Knowledge Base into DeepSeek Harness

doco-dsh is the native plugin for DeepSeek Harness, designed to connect a Doco knowledge base and address the complexity of external Agent integration. The plugin provides nine tools that support block-level addressing, full-text search, structured reading, and draft saving. Write operations are disabled by default and require approval and scope validation. The plugin depends on the host environment, with API and permissions configured through environment variables, and can be installed with one click via the plugin manager. Typical usage includes authorizing access with `/doco connect`, retrieving spreadsheet structure, and performing atomic batch data updates.

Read More
dsh-prompt-optimizer: One-Click Prompt Rewriting

This plugin is designed to address the issue of unclear prompt writing in DeepSeek Harness. By integrating a “✨ Optimize Prompt” button beneath the editor, it leverages an LLM to perform a six-step structured workflow (intent, context, constraints, and more) that rewrites rough drafts into high-quality prompts. The plugin supports one-click replacement into the input box, preserves the original language and code, and is easy to install, effectively lowering the barrier to authoring high-quality prompts.

Read More
dsh-token-usage-dashboard: DeepSeek Harness Token Usage Dashboard

DeepSeek Harness (DSH) has launched the `dsh-token-usage-dashboard` plugin, providing developers with real-time Token monitoring. Through a GitHub-style heatmap and metric cards, it aggregates key indicators such as total tokens, daily peak, and active sessions. It supports daily/weekly views, model rankings, and real-time data synchronization. To install it, run `dsh plugin --profile web add github:solstice621/dsh-token-usage-dashboard`. After enabling the plugin, you can view the dashboard in Settings → Stats. Please note potential plugin conflicts, environment dependencies, and data statistics rules.

Read More
DeepSeek Harness Plugin: Alibaba Cloud Bailian Kimi Model Adapter

`dsh-llm-bailian-kimi` is a DeepSeek Harness plugin designed to address issues that arise when integrating with Alibaba Cloud Bailian, such as the native adapter’s lack of support for image input and system-prompt errors. By adapting to the OpenAI-compatible interface, it enables multimodal support, chain-of-thought reasoning, tool calling, and zero-build installation for Kimi models. When using it, be aware of Bailian-specific limitations, including the requirement to send system prompts with the `system` role, using string-based checks for tool-call continuations, and configuring streaming timeouts.

Read More