Every DeepSeek Harness plugin

What it is — DeepSeek Harness multi-model routing plugin that binds custom professional agents for task routing.

Who it is for — If you need to handle vision or translation tasks with separate models, install this plugin. If you do not use multi-modal account pools, you can skip this plugin.

Watch out — Sandbox testing passed. It works in a new profile and is registered by harness. No obvious pitfalls found.

The verdict — I would install it because it supports binding professional agents, but only if you accept multi-model account configuration.

What it is — dsh-web-tools is a DeepSeek Harness plugin that integrates multi-provider web search and fetch, supporting Tavily, Exa, Firecrawl and others with native settings UI.

Who it is for — Users who need to call web_search and web_fetch tools during DeepSeek Harness responses are suitable for this plugin. For users who only need a single provider, this plugin is not necessary.

Watch out — Sandbox testing passed: the plugin was installed in a fresh profile and successfully registered as web.searchProvider. No obvious pitfalls found.

The verdict — I would install it because it provides quota display and API key health status checks.

What it is — Sync Wallpaper Engine current wallpaper to DeepSeek Harness web UI frosted glass background, display-only.

Who it is for — When running in DeepSeek Harness web profile with Wallpaper Engine installed, the scenario requiring wallpaper sync to page background. If your main need is direct desktop wallpaper modification, please operate it inside Wallpaper Engine, you don't need this plugin.

Watch out — Sandbox test passed in new profile, registered into profile by harness. No obvious pitfalls found. The panel transparency, blur, and shadow slider values are saved only in page memory and reset on refresh.

The verdict — I would install it because it only reads WE state without modifying desktop wallpaper and sandbox test passed.

What it is — dsh-whale-meter plugin tracks token consumption in DeepSeek Harness sessions.

Who it is for — Users who conduct sessions on DeepSeek Harness and want to monitor token tiers are suitable to install it. Users who avoid token cost analysis can use other tools without monitoring.

Watch out — Sandbox testing shows it requires manual approval for the build to register successfully. npm one-liner does not require this step.

The verdict — I won't install it because sandbox testing requires manual approval for the build to complete registration, which is not a one-command process.

What it is — A Git GUI plugin for DeepSeek Harness Web UI, enabling viewing, staging, committing, and reverting project changes in the browser while seeing real-time Agent modifications to workspace files.

Who it is for — When you rely on DeepSeek Harness Agent to modify project files and want real-time visibility of those changes in the browser, this plugin is suitable. If your workflow does not involve Agent-automated file editing or integration of Git management into a web interface, the plugin's functionality is not directly relevant.

Watch out — Sandbox test shows it can be installed but the build script is blocked by pnpm. Users need to manually allow the build script from source installation to complete registration. No obvious issues found.

The verdict — I would install it because it enables real-time visibility of Agent file modifications in the browser, but I would need to manually allow the build script when installing from source.

What it is — minecraft-dev plugin injects Minecraft development skills, tools and expert preset into DeepSeek Harness.

Who it is for — When you use DeepSeek Harness to develop Paper plugins, this plugin can provide Java build knowledge and API references. If your work only involves non-Java programming language application development, you do not need this plugin.

Watch out — Sandbox test shows successful installation and registration in a new profile. No obvious pitfalls found. Static check indicates it can execute shell commands.

The verdict — I would install it because it includes a sub-agent team to assist in serial completion of plugin development stages.

What it is — The plugin pops up a panel to remind backing up chat session logs when opening the DeepSeek Harness web interface.

Who it is for — It is suitable for users needing to sync chat records in cross-device office work. It is not suitable for users who prefer manual management of session logs on a single device.

Watch out — Sandbox test passed after installation in a new profile and registration. Runtime settings are persisted via popup. No obvious compatibility issues.

The verdict — I would install it because it supports format switching for different needs.

What it is — Xiao Liu Ren divination plugin for DeepSeek Harness: casts using time, numbers or random seed, returns conclusion-first plain-language verdict plus six-palace layout.

Who it is for — Users of DeepSeek Harness who need reproducible divination or traditional cultural references for decision-making scenarios. Not suitable for users who do not use divination tools or prefer not to execute shell commands.

Watch out — Installation requires GitHub source with manual build script approval. Sandbox test passed, successfully registered in new profile, but full restart of harness web is needed for widget to appear. No obvious issues found.

The verdict — Will install it because of zero-dependency core and conclusion-first output, useful for integrating traditional references.

What it is — This plugin adds a collapsible iOS-style stats panel under the composer to display session costs, account balance, cache-hit rate, and token usage.

Who it is for — When you are using DeepSeek models for multi-turn conversations and need to monitor token usage and account balance, this plugin is suitable for installation. When you are running single-turn queries that do not involve token tracking, this plugin does not match your needs.

Watch out — Sandbox testing passed: the plugin was installed in a new profile and registered by Harness. Display functions normally. No obvious issues detected.

The verdict — I would install it because it fetches the real-time balance from the user's own DeepSeek API key without any additional configuration.

What it is — The adhd_breakdown tool scans tasks and provides structured next steps.

Who it is for — Users who want to fill the behavioral and mental health gap in the dsh ecosystem can install this plugin. Users who are not in the behavioral and mental health category may find it unsuitable.

Watch out — Sandbox testing passed: installed successfully in a new profile and registered by harness. Static check indicates possible shell command execution. No obvious issues found.

The verdict — I would not install it because installing from source requires manually allowing its build script.

What it is — This plugin registers four faithful palettes from Solarized and Selenized into the DeepSeek Harness Web theme runtime.

Who it is for — If you are using DeepSeek Harness Web and like Solarized or Selenized color schemes, this plugin is suitable for you. If you only use DeepSeek Harness in a command-line environment, you do not need this plugin.

Watch out — The sandbox test passed: it was successfully installed in a new profile and registered with Harness. No obvious issues found.

The verdict — I would install it provided that I am using DeepSeek Harness Web.

What it is — dsh-pet is a desktop pet plugin supporting sprite animation, agent state linkage, drag, and reminder tools.

Who it is for — If you use the DeepSeek Harness agent for complex tasks and need visual cues to stay focused, dsh-pet's state linkage suits you. Users who complete work only via command line or text interface do not need it because they require no extra reminders.

Watch out — Tested successfully in a clean profile where it registers with harness. No obvious issues found; the settings panel supports adjusting size, position, and skin preview.

The verdict — I would install it because it provides visual feedback during agent thinking, but only if your workflow needs timed reminders or state cues.

What it is — dsh-vision-bridge allows text-only models to see images in the input box.

Who it is for — If you are using DeepSeek V4 for text conversations needing image processing, dsh-vision-bridge suits you. If you are using a vision-native model or avoid image content, you do not need this plugin.

Watch out — The install command succeeds. No obvious issues found. Configuration requires an OpenAI-compatible multimodal model, such as qwen-vl-max.

The verdict — I would install it because it enables text models to see images in conversations, but sandbox testing shows it loads as ordinary dependency without extending the profile layer, so use it per the README.

What it is — Tool for backing up and restoring DeepSeek Harness user data with scheduled auto-backup and GitHub sync.

Who it is for — Users who need to back up and sync DeepSeek Harness user data across macOS, Linux, and Windows to a private GitHub repository. Users who do not have tar in their PATH are not suitable because the tool requires it.

Watch out — Sandbox testing passed and it can be registered in a new profile. Installation from source requires manually approving the build script. Backups contain plaintext credentials, so do not sync to untrusted locations.

The verdict — I would install it because it provides path-traversal rejection during restore and GitHub sync support, but only if using a private repository.

What it is — It is the host-side Typert remote for dsh-tool-writing.

Who it is for — People using the DSH writing panel and relying on dsh-tool-writing engine are suitable. If you use only client-side running writing tools, you don't need this plugin.

Watch out — Sandbox test passed, successfully registered in fresh profile. Requires dsh-tool-writing@0.3.0 dependency. No obvious issues found.

The verdict — I wouldn't install it because it only exposes read data and write operations remain on the host side engine.

What it is — Client-side writing panel and full-book index view for DeepSeek Harness, shown only in writing-preset sessions, covering project volumes, stats, corpus library, full-text search, and evolution graph.

Who it is for — Users performing long-form novel writing with the writing preset benefit from this plugin for sidebar panel and index view support. Users not engaged in novel writing or evolution tracking can skip this plugin.

Watch out — Tested successfully in a fresh profile where it registers with Harness. Requires host-side installation of related writing services. No obvious issues found.

The verdict — I would install it because it enhances writing workflow with data management and search, but only if host-side writing tools are already set up.

What it is — KnowLP-RAG plugin provides dual knowledge-graph RAG for Markdown notes with P/S-Agent traversal and weight feedback loop.

Who it is for — When using DeepSeek Harness to process large Markdown notes that require retrieval with dependency chains and similarity alternatives, this plugin is suitable to install. Users who only need simple vector or full-text search with small note volume need not install it.

Watch out — After install, set KNOWLP_VAULT and KNOWLP_GRAPH_DIR environment variables manually. Sandbox test passed in fresh profile with registration, no obvious issues found.

The verdict — I would not install it because it requires Python environment bootstrapping and is only useful in specific note scenarios.

What it is — Removes the folder selection requirement before starting a conversation in the DeepSeek Harness Web interface.

Who it is for — When conducting conversations in DeepSeek Harness Web, the goal is to avoid manually selecting a workspace. Users who are accustomed to manually selecting workspaces can handle the folder choice through alternative approaches.

Watch out — Sandbox test passed with successful installation and registration into profile. No obvious issues found. The static check indicates execution of shell commands.

The verdict — I would install it because it handles zero-workspace startup by automatically creating the default directory and opening the input box, but only if the default directory suits my daily workflow.

What it is — dsh-opencode-go-usage plugin displays OpenCode Go subscription usage in the DeepSeek Harness composer dock with rolling/weekly/monthly windows and reset countdowns, including a built-in credential editor.

Who it is for — If you use OpenCode Go models in web interface management and need to monitor subscription remaining time and countdowns, this plugin fits.

Users not using OpenCode Go don't need to install it for this purpose.

Watch out — Sandbox test passed: installed in new profile and registered by harness. Static check shows it executes shell commands. Requires OpenCode Go session cookie treated like password. No obvious pitfalls found; from source install, manual build script approval needed if build blocked.

The verdict — I wouldn't install it because it only monitors OpenCode Go usage and requires a full session cookie, which may not justify the effort unless you frequently need web-based stats.

What it is — dsh-s1 provides first-class s1_* tools for web research in DeepSeek Harness.

Who it is for — When using DeepSeek Harness for web research, this plugin can provide s1_* tool support. For users not running in a profile environment, it cannot be used.

Watch out — Sandbox test passed in a new profile. It requires S1_KEY to function. No obvious issues found.

The verdict — I would install it because it runs in-process with a bundled s1 skill.

What it is — Uses a vision sidecar to describe image attachments, enabling image paste on text-only DeepSeek models.

Who it is for — Use it when running official text-only models in the web UI and pasting screenshots to query red text. Users who prefer native vision models don't need this plugin.

Watch out — Sandbox test shows pnpm blocks the build script for github:shinjiyu/dsh-plugin-multimodal, requiring manual allowance for registration. Tests found no other obvious issues.

The verdict — I would install it because it enables image paste on text-only models but requires manual build approval.

What it is — Optimize raw instruction into professional four-section prompt (Role / Task / Context / Format) via Harness llm service, with composer button and auto-optimize hook.

Who it is for — When building complex agents requiring structured prompts using DeepSeek Harness, this plugin provides the prompt_optimize tool and ctx.promptOptimizer service. Users who prefer raw input formats can utilize its composer button and hook for optimization options.

Watch out — Sandbox test passed: installed in new profile and registered by harness. No obvious pits found. Supports configurable options like default temperature 0.2 and maxTokens 1200.

The verdict — I would install it if the agent needs structured prompts because it provides tools and services to reduce manual optimization work, assuming integration with DeepSeek Harness.

What it is — Adds support for the nono sandbox backend.

Who it is for — If you need to use the nono sandbox backend to execute tasks in DeepSeek Harness, install this plugin. Windows users do not need to install this plugin because they lack native binding support.

Watch out — Real test shows it can be installed but the build script github:omdsh-dev/sandbox-nono is blocked by pnpm, preventing harness from completing registration. Users need to manually allow that build to install properly. No other obvious issues found.

The verdict — If I were you I would go for installing it since it enables the distinct sandbox-nono row via profile overlay for nono backend use, yet from source install one must manually permit the build script.