Every DeepSeek Harness plugin

What it is — Packages DeepSeek Harness as a native desktop application.

Who it is for — Users who use DeepSeek Harness on macOS. Users who run DeepSeek Harness in the browser.

Watch out — Sandbox test passed: it registers in harness after installation in a fresh profile. Source installation requires Node.js, pnpm, Rust and Tauri native prerequisites. Windows version does not support persistent Bash sessions.

The verdict — I will install it because it provides native desktop support, but only if I accept unsigned packages.

What it is — Embeds DeepSeek Harness natively into the Electron main process as a Windows desktop application.

Who it is for — For Windows users who want a GUI interface for DeepSeek Harness without needing Node setup. For users who operate primarily on macOS or Linux with terminal workflows, this does not apply.

Watch out — Installation may automatically execute scripts. Source installation requires manually approving the build script. Sandbox test passed with registration into harness in a fresh profile. No obvious runtime issues found.

The verdict — I would install it because it provides native embedding that addresses an upstream session-log corruption bug.

What it is — Provides a desktop version of DeepSeek Harness

Who it is for — Suitable for users who want to use the built-in plugin market. TypeScript users don't need to install it.

Watch out — Installation requires sourcing from GitHub. No obvious pit found.

The verdict — I wouldn't install it because it's a TypeScript project.

What it is — A Windows plugin for DeepSeek Harness that provides update checking, server control, system tray, and graphical console.

Who it is for — Windows users running DeepSeek Harness who need a system tray for quick server start/stop and a graphical console for management are suitable for installation.

Users who only run the official Web version without needing PowerShell scripts or WPF interfaces can skip this plugin.

Watch out — Installation requires manual permission for the build script from GitHub source. Tested successfully in a fresh profile where it registers and loads into harness. No other issues found.

The verdict — I would install it because it offers PowerShell-based process management and non-automatic update checks, but only on Windows where such reminders are needed.

What it is — Harness Hub provides a desktop client and plugin marketplace for DeepSeek Harness.

Who it is for — If you want to use the DeepSeek Harness desktop client for agent tasks, this tool is suitable. If you prefer the command line dsh environment without needing desktop service hosting, you can continue using command line dsh.

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

The verdict — Installing Harness Hub is worthwhile for its automatic profile backup and rollback, but only when using the DeepSeek Harness desktop client.

What it is — GitLearnOS provides a native DeepSeek Harness panel for Git learning with learner-owned memory.

Who it is for — If you use DeepSeek Harness to manage AI learning where the main agent needs learner-owned Git memory, install it. If you do not use the Harness panel or do not want the AI agent to access Git as evidence, you do not need it.

Watch out — Sandbox test passed: installed in a fresh profile and registered into Harness. Installation from source requires manually approving the build script. No obvious issues found.

The verdict — I would install it because it enables the main agent to directly own and commit Git learning transactions within Harness.

What it is — DeepSeek Harness skill for Chinese official document format checking and content optimization based on GB/T 9704.

Who it is for — When drafting or reviewing official Chinese government documents using DeepSeek models that require strict format compliance and content optimization, this skill is suitable. Users not using DeepSeek Harness or handling non-Chinese official documents do not need it.

Watch out — Sandbox test passed: successfully installed and registered in a fresh profile. Executes shell commands. No other issues found.

The verdict — I would install it because it provides native integration with DeepSeek Harness for official document processing, offering value in compliant Chinese document workflows, though it requires shell execution capability.

What it is — SmokingGun is a static scanning tool that provides optimization evidence for agents by finding complexity hotspots.

Who it is for — When using DeepSeek Harness to help agents find code complexity hotspots, this tool can generate reports including hotspots, coverage, and assumptions. If your agent host does not support Node 22.18 or higher, you do not need it.

Watch out — It installs successfully as an ordinary dependency because its package.json lacks dsh.bundle, so it does not become a profile layer. This is not an installation failure; it simply is not a bundle-type plugin. No obvious issues found.

The verdict — I would install it if you need optimization evidence for agents because it can let agents decide the next investigation step, but you must use it as described in the README and it will not add a profile layer.

What it is — Builds wave-style RAG detail retrieval for Zotero paper libraries by porting VCPToolBox semantic dynamics ideas.

Who it is for — Researchers using DeepSeek Harness for Zotero paper literature studies needing diverse results and claim-evidence fidelity check can install it. If not using Zotero data directory or not requiring claim-evidence fidelity check, can skip installation.

Watch out — Installation requires sourcing from GitHub. Sandbox test passed in fresh profile. No obvious issues found.

The verdict — I would install it because it offers unique topological retrieval for Zotero workflows, provided I manage DSH patch files after updates.

What it is — dsh-revive plugin scans persistent sessions after DSH restart and sends continue instructions to revive interrupted sessions.

Who it is for — Users running long tasks on DSH that are prone to crashes should install this plugin to automatically revive sessions. Users without persistent storage needs should not install it since scanning is unnecessary.

Watch out — Installation from source requires manually allowing the build script to complete registration. Sandbox test shows it works with no extra configuration. No obvious issues found.

The verdict — I would install it in DSH with high crash risk, provided manual build permission is granted.

What it is — dsh-skins is a client plugin for DeepSeek Harness Web GUI that registers skins via the official ThemeService, supporting nord token-based and aurora image-based skin switching.

Who it is for — A user running DSH in dark mode who wants to match the nord color scheme can install this plugin. Users who prefer a fixed UI appearance without skin switching options will not benefit from this plugin.

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

The verdict — I would install it because it provides multiple skin options for the DSH web interface without requiring core modifications.

What it is — dsh-companion is the companion mode plugin for DeepSeek Harness, providing SOUL personality, Hermes-style long-term memory, and optional QQ channel.

Who it is for — For users performing single-user companionship conversations with DeepSeek Harness, this plugin is suitable to supplement personality and long-term memory capabilities. Users needing multi-chat targets may find it unsuitable because it is a single-target MVP.

Watch out — No obvious issues found. Sandbox testing passed in a new profile and was registered by harness. Configuration requires adding memoriesDir and optionally enabling the QQ channel.

The verdict — I would install it because it can directly reuse the selfloom SOUL.md / USER.md / MEMORY.md data source.

What it is — This is a declarative deny-by-default tool policy plugin that integrates at DeepSeek Harness tools/pre-execute. It supports denying tool calls, routing to human approval, or proxying unchanged.

Who it is for — Install this plugin if you run unattended jobs with DeepSeek Harness models and need control over tool call policy. It is not for users who rely solely on Harness built-in sandboxing without extra policy routing.

Watch out — GitHub source install is blocked by pnpm on the prepare script for this repo; manual build allow is required before Harness registers it. Static check shows it executes shell commands; no other issues found.

The verdict — I would install it because it adds declarative deny-by-default policy routing, but only after manually whitelisting the build script; built-in Harness mechanisms are sufficient otherwise.

What it is — A model-native meta-agent plugin that lets the model synthesize task-specific workflows at runtime to coordinate tools.

Who it is for — Users running multi-step tasks in DeepSeek Harness who want the plan to be synthesized by the model at runtime rather than implicit are suitable for this plugin. Users preferring fixed agent topologies may find this plugin offers a different approach to dynamic pattern selection.

Watch out — Successfully installed and registered in a new profile with no obvious issues found. This plugin records plans but does not execute any work, so there are no additional runtime dependencies. Manual approval of the build script is required when installing from source.

The verdict — I would install it because it enables the model to decide the workflow at runtime and keeps plans versioned and durable.

What it is — Displays balance and Coding Plan quota for configured platforms in the bottom right of DeepSeek Harness web profile.

Who it is for — Users monitoring balance and Coding Plan quota across multiple platforms using API keys in DeepSeek Harness web profile. Headless profile users do not need to install this plugin.

Watch out — Sandbox test passed. It was placed in a fresh profile and registered into the profile by DeepSeek Harness.

The verdict — I would install it because it can automatically identify and query configured platforms for balance and quota, but only if you have at least one API key configured and are using a web profile.

What it is — The dsh-safe-delete plugin moves files to trash for safe deletion, with restore and purge support.

Who it is for — When using DeepSeek Harness to perform delete operations, this plugin is suitable to install. For users who mainly handle remote commands, this plugin has limited effect on your use case.

Watch out — Sandbox test shows successful installation, but the package.json lacks dsh.bundle so DSH treats it as ordinary dependency rather than profile layer. This is not installation failure because the plugin is not a bundle type. No obvious pitfalls were found.

The verdict — I would install it because it provides a safe deletion alternative and guides use of safe_delete, but the premise is that it installs as ordinary dependency and must be used per the README instructions.

What it is — It runs Pi extensions inside DeepSeek Harness through a compatibility host.

Who it is for — When using DeepSeek Harness to run Pi extensions in a DSH agent, this plugin is suitable. It is not for users who want direct source integration of Pi extensions into DeepSeek Harness.

Watch out — Sandbox testing shows it can be installed via npm, but pnpm blocks the build scripts for @google/genai@1.52.0 and protobufjs@7.6.5, preventing Harness from completing registration. Manual allowance of the build is required to fully install it. No other obvious issues found.

The verdict — I would install it because it enables running Pi extensions in DSH, but the build must be manually allowed as a prerequisite.

What it is — This is a DeepSeek Harness plugin providing policy-enforced and evidence-first governed workflows.

Who it is for — Users requiring policy-enforced governance for DeepSeek Harness agents are suitable for installing this plugin. It does not suit users who do not need evidence recording or non-bypassable runtime boundaries.

Watch out — Sandbox tests indicate that pnpm blocks a build script, requiring manual allowance of the GitHub build to complete registration. After installation, persistent reload is refused due to upstream DSH limitation. No other obvious issues found.

The verdict — I would not install it because it remains a developer preview with pending independent review.

What it is — Integrate default agent preset switching into the new session button.

Who it is for — When using DeepSeek Harness to create new session, when switching agent preset. Users who use running sessions are not suitable to install, because the plugin only affects the current blank session.

Watch out — Sandbox test passed, installed in new profile and registered into profile. No obvious pit found.

The verdict — I would install it because it is a DeepSeek Harness profile plugin.