Every DeepSeek Harness plugin

What it is — This plugin displays conversation progress and token usage for each session, offering all-window overview, session analysis, cost estimation, and token heatmap.

Who it is for — When using DeepSeek Harness to manage multiple sessions and need to view progress and token usage across sessions, this plugin is suitable.
If you only work within a single session without needing cross-session management, you can skip this plugin because its core adds global-window observability.

Watch out — No obvious issues found. Sandbox test passed: installed in a fresh profile and automatically registered by harness.

The verdict — I would install it because it provides a local, read-only way to observe all sessions without external requests or session writes.

What it is — dsh-role-router implements role-based model routing in DeepSeek Harness, forcing planner role to configured models in plan mode and supporting subagent, with Web UI settings card and composer summary.

Who it is for — It suits users who want to force a specific model for the planner role during plan mode in DeepSeek Harness; those who do not leverage DeepSeek Harness's composer or multi-role routing can skip it.

Watch out — Sandbox testing confirmed successful installation and registration. Unregistered providers for planner or subagent roles result in NO_ADAPTER errors. No other issues found.

The verdict — I would install it because it enables explicit model selection per role in DeepSeek Harness, which simplifies workflows involving plan mode and composer interactions.

What it is — dsh-plugin-vision plugin for DeepSeek Harness adds image description, OCR, and VQA tools to text models via free Gemini and GLM vision APIs.

Who it is for — If your DeepSeek Harness text model needs to analyze screenshots or documents, install dsh-plugin-vision. Users who only use pure text do not need it, unless they require visual input.

Watch out — Sandbox testing passed on a fresh profile with registration. Static checks indicate shell command execution. Costs rely on free Gemini and GLM vision APIs, with negligible plugin token overhead.

The verdict — I would install it because it enables visual analysis for text models via free APIs, but only if you need to handle images.

What it is — dsh-home-ui is a pure frontend plugin for visual refinement of home feed information flow on the DeepSeek Harness web client.

Who it is for — If you browse home feed information flow on the DeepSeek Harness web client and want unified sidebar and feed coloring, install this plugin. If you use it on mobile without needing wide screen mode, you don't need to install this plugin.

Watch out — Sandbox test passed in a fresh profile after registration with harness. It is a pure frontend extension with no core changes. No obvious pitfalls were found.

The verdict — Switching to my perspective, I would install it since it's a pure frontend extension with no core changes and passes the sandbox test.

What it is — A tool to check protocol issues in DeepSeek tool calling, including reasoning_content, strict schemas, and captured SSE streams.

Who it is for — When developing agents using DeepSeek models with tool calling and needing to debug protocol problems in requests and message history, this plugin can help. If your work does not involve DeepSeek agent tool calling flows, this utility is irrelevant.

Watch out — Sandbox testing passed, with successful registration in a fresh profile. No obvious issues detected. It requires a Python 3.10+ environment to run.

The verdict — I would install it because it helps identify specific issues in DeepSeek tool calling protocols like reasoning_content, but only if my development involves complex tool loops.

What it is — DeepSeek Harness local data unified dashboard and AI steward tool to manage conversations, skills, and other data origins and locations.

Who it is for — Users who need to track the origins and locations of conversations and skills in DeepSeek Harness are suitable for installing it. Users who primarily use command-line tools can skip it because it mainly enhances the web interface.

Watch out — Sandbox test passed: installed in new profile and registered by harness. No obvious pitfalls found.

The verdict — I would install it because it provides AI tools for data inspection, but only if using the web profile.

What it is — dsh-content-studio is a DeepSeek Harness bundle plugin providing integrated tools for screenshots, screen recordings, Markdown to Xiaohongshu image cards, dev.to and Medium publishing, and human-in-the-loop review.

Who it is for — It suits users creating Xiaohongshu posts on macOS with DeepSeek Harness who need text-to-image generation and automated publishing.
Users without macOS screenshot and recording needs can skip it.

Watch out — Sandbox testing confirmed successful registration in a fresh profile. Dependencies include macOS, ffmpeg, and Google Chrome. API keys are required for image generation and dev.to publishing. Cross-machine setup requires adjusting the xhs-mcp Chrome executable path.

The verdict — I would install it if I need to bundle screenshots, recordings, card rendering and publishing tools together for macOS-based content workflows.

What it is — The plugin displays a 365-day daily token usage heatmap for DeepSeek Harness with hover details on models and token amounts.

Who it is for — Suitable for users who need to track token consumption when using multiple models in DeepSeek Harness sessions. Those who only care about model output without looking at token consumption do not need to install it.

Watch out — Sandbox test passed with installation and registration successful in a new profile. Historical sessions are automatically backfilled with real-time progress display. Statistics may differ from the official platform due to auxiliary model calls and other factors.

The verdict — I would install it because it provides a token usage heatmap based on session logs, but only if I need to monitor model consumption rather than direct billing.

What it is — DeepSeek Harness Desktop is a plugin that converts the Web UI into a native Tauri desktop application.

Who it is for — Users who want to run DeepSeek Harness in a new profile with native window support are suitable for installing it. Users who prefer Electron desktop solutions can consider other desktop shells.

Watch out — Sandbox test passed: installed in a new profile and registered with the harness. No obvious issues found. It is known that v1 lacks runtime hot update.

The verdict — I would install it because it supports background automatic updates with rollback.

What it is — DeepSeek Harness Core is a self-evolving AI persona core plugin for DeepSeek Harness, including autonomous iteration of heart/emotion/body/brain/soul systems and a desktop pet.

Who it is for — If you are using DeepSeek Harness to build AI persona systems that support multi-tier memory accumulation and autonomous evolution, this plugin provides iterative capabilities for the five organ modules. If you prefer completely fixed presets, you don't need to install it because this plugin will automatically evolve.

Watch out — Sandbox test shows that after source installation, harness does not register the plugin into the profile. No other obvious issues found.

The verdict — If the profile registration issue is resolved, I would consider installing it because it provides self-evolving personality capabilities, but the installation is currently incomplete.

What it is — Adds an entry to the session header for opening the current working directory in a preferred editor.

Who it is for — When performing code development in DeepSeek Harness and needing to quickly open the current project directory. If you only need to complete all session operations within the browser without external tools.

Watch out — Sandbox testing passed: it was successfully registered in a new profile. Shell command execution is required for normal plugin operation. No other issues discovered.

The verdict — I would not install it because it requires installation from source and it will execute shell commands.

What it is — DeepSeek Harness plugin provides reload_plugin tool to restart matched Cordis Loader entry.

Who it is for — When you need to update mcp-client plugin code, this plugin fits you. This plugin does not apply if your profile does not use web or headless bundle.

Watch out — Sandbox test passed: the plugin was installed in a fresh profile and registered by the harness. No obvious issues found.

The verdict — I would install it because it supports reloading without restarting the host, but only if using the web bundle.

What it is — DSH plugin management service. Provides install, rollback, and audit operations as a service.

Who it is for — When using hotplug tools in agent sessions, this plugin is suitable. It provides the capability to manage plugins in DSH. If you do not need to call these tools, you do not need to install it.

Watch out — Tested successfully in a new profile, registered into harness after install. Executes shell commands. Manual approval for build script may be needed when installing from source. No other issues found.

The verdict — I would install it if managing plugins in agent sessions is required, because it encapsulates DSH's plugin operation chain.

What it is — DSH plugin that converts any source text or session trace into a knowledge graph with bidirectional linking to the original text.

Who it is for — When using DeepSeek Harness in a session and wanting to convert selected chat text into a knowledge graph to aid content understanding. If the plugin is not designed for session trajectory splitting, you can skip it.

Watch out — When installing from source, you need to manually allow the build script to run first. Sandbox test passed: successfully installed in a new profile and registered in the profile. No obvious issues found.

The verdict — I would not install it because the static check indicates it executes shell commands, which adds usage difficulty.

What it is — DSH Plugins 4U integrates WeChat session bridging, preset wallpapers and image analysis via source install into DeepSeek Harness.

Who it is for — If you need bidirectional WeChat-DSH session sync for chats with DeepSeek models, this collection suits. Users who rely solely on built-in or local wallpapers need no extra integration.

Watch out — Requires Git and Node.js 20+, runs shell commands for bridging. Image analysis needs SiliconFlow API Key. Sandbox test passed on fresh profile with registration, no obvious issues found.

The verdict — I would install it for the multi-feature integration, but only if you are already running DSH and can handle API Key setup plus WeChat QR scan.

What it is — loongport-dsh integrates LoongPort OpenAI-compatible providers into DeepSeek Harness and provides a CLI for custom endpoint configuration.

Who it is for — If you need to connect to LoongPort verified providers through DeepSeek Harness, you can install this plugin. If you prefer to avoid command-line tools entirely, this plugin may not fit since it includes CLI functionality.

Watch out — Sandbox testing passed after installation into a new profile with successful registration. No obvious issues were found. However, the plugin executes shell commands.

The verdict — I would install it because it allows batch registration of multiple models via CLI, but with the premise that API keys must be configured manually.

What it is — This plugin integrates Qwen-MM-Plugins into DeepSeek Harness, supplying 12 multimodal MCP tools and a web settings interface.

Who it is for — If you use a multimodal model to analyze images or audio files, install this plugin. It enables the model to call tools such as vision_chat and ocr.

Watch out — Manual approval of the build script is required for GitHub source installation. Sandbox testing passed: the plugin installs and registers in a fresh profile. A Qwen API key is needed to activate tool functions.

The verdict — I will not install it, because the static check flags shell command execution and it requires a Qwen API key.

What it is — This is a plugin bridging the Feishu bot to DeepSeek Harness.

Who it is for — Users that need to receive messages through Feishu and execute tasks using DeepSeek Harness are suitable to install this plugin. Users who do not involve Feishu message passing can choose not to install this plugin.

Watch out — Sandbox test passed. This plugin executes shell commands and requires providing an API key or token. No obvious pitfalls found.

The verdict — I would install it because it supports file transfer.

What it is — It is a terminal-native coding agent built on DeepSeek Harness, supporting Minimal/Router profiles and Windows platform.

Who it is for — It is suitable for installation when performing coding tasks in a project directory using DeepSeek Harness plugin with Minimal or Router profiles. Those who do not need resumable sessions or persistent TUI do not need to install it.

Watch out — The sandbox test passed when installed in a new profile and registered with harness. Installation from source requires executing the build script and manually allowing it. No obvious pitfalls found.

The verdict — I will install it because it supports MCP and continuable subagents in a terminal environment, but the premise is using compatible terminals like Windows Terminal and manually allowing the build script.

What it is — dsh-plugin-manager adds a "plugins" button to the sidebar in DeepSeek Harness web GUI for viewing and managing user-installed plugins.

Who it is for — If you need a sidebar in the web GUI to manage the status and switches of installed plugins, this plugin allows unified application of changes. If you only want to install individual plugins without needing batch management, you don't need to install this.

Watch out — It was tested to register successfully in a new profile. After installation, restart dsh web for the plugin to mount. No obvious issues found.

The verdict — I would install it because it enables discovery of community plugins via awesome-dsh-plugin and unified control, but only if you are okay with restarting the web interface afterward.

What it is — This plugin is the Feishu private-chat frontend for DeepSeek Harness.

Who it is for — If you need to manage DeepSeek Harness agents through Feishu private chats, this plugin is suitable for your scenario. If your agent management is done via the web GUI without Feishu, you may not need this plugin.

Watch out — Sandbox test passed: installed into a new profile and registered by Harness. No obvious issues found.

The verdict — I would not install it because the web GUI already shares sessions.

What it is — DeepSeek Harness plugin that predicts sandbox operation risks with Flash model, auto-approving repeatable actions and routing risky ones to human review.

Who it is for — When running automated tasks in DeepSeek Harness sandbox environment using Flash model, the plugin helps by predicting if confirmation is needed. Users who do not perform sandboxed operations or do not use Flash for risk prediction do not need this plugin.

Watch out — Configuration involves adding auto-approve preset to ~/.dsh/profiles/web/cordis.patch.yml, followed by restarting dsh web and selecting the preset. Sandbox testing succeeded after installing in a fresh profile and registering the plugin with harness. No obvious issues detected.

The verdict — I would install it to reduce manual approvals for sandbox tasks via Flash prediction, provided the cordis.patch.yml preset is configured first.

What it is — dsh-llm-proxy is a global HTTP proxy plugin for DeepSeek Harness. It uses undici's setGlobalDispatcher and EnvHttpProxyAgent to intercept LLM request routing with configuration, hot switching and API observability.

Who it is for — When you need LLM requests in DeepSeek Harness to go through a proxy to access external endpoints, this plugin can take over the fetch routing. If your LLM endpoint can connect directly, you do not need this plugin.

Watch out — Sandbox testing passed, it registers and activates normally in a fresh profile. Installation from source requires manually allowing the build script. The web settings page requires manually adding the llm-proxy namespace to the whitelist to show the form.

The verdict — This plugin is worth installing when your LLM endpoints require proxy access because it enables configuration and hot switching for dsh's fetch routing, but the value depends on whether you need these capabilities.

What it is — omdp is a DeepSeek Harness plugin monorepo that bundles dsh-connector, dsh-vision-bridge, and dsh-gitbash-win as independently installable packages.

Who it is for — If you use text-only models and need vision capabilities, or need POSIX shell tools on Windows, or want to centrally manage MCP servers and user skills, you can install omdp.

If your work only uses native tools in Linux or macOS environments, or the model already supports multimodal, and does not need these tools, you can skip installing it.

Watch out — The installation process will execute shell commands. Sandbox testing passed: after installing in a fresh profile, it is registered into the profile by harness. No other issues found.

The verdict — If your work requires simultaneously managing multiple DeepSeek Harness plugins, I would install it, because it packages them into a monorepo for centralized use.