Every DeepSeek Harness plugin

What it is — This is a multi-perspective structural analysis plugin for complex social, economic, philosophical questions.

Who it is for — When you need structural decomposition for complex social economic issues in DeepSeek Harness, this plugin provides multi-angle support. For users who only need quick summaries or information queries, this plugin does not match their needs.

Watch out — Sandbox test passed and successfully registered in a new profile. No obvious pitfalls found.

The verdict — I would install it because it uses an attack-loop protocol against analytical priors for more reliable conclusions.

What it is — Plugin that converts Markdown to academic-formatted Word documents.

Who it is for — When writing academic papers that require Markdown to Word conversion, people who need the formatting can install this plugin. If your work does not involve LaTeX formulas or tables, this plugin cannot provide the needed features.

Watch out — Sandbox test shows it installs successfully but is treated as ordinary dependency rather than profile layer. Python dependencies are managed automatically in a dedicated virtual environment. No obvious issues found.

The verdict — I would install it because it automates academic formatting from Markdown, but only if my work involves such conversion.

What it is — Python CLI tool to repair installation issues for AI coding agents.

Who it is for — If you encounter postinstall script not running with agents like Claude Code or OpenCode during coding, this tool can diagnose and fix the issues. If your agent installation is stable in a local Node.js environment without frequent upgrades, you can use more targeted debugging tools.

Watch out — Installation from GitHub source requires manually allowing the build script to run. Sandbox test passed.

The verdict — I will install it because it supports self-healing on agent startup and DeepSeek Harness repair, but only if I need cross-platform fixes for multiple agents.

What it is — It is a CLI-first vision sidecar that enables text-only coding agents to analyze screenshots, diagrams, charts, UI diffs, and videos using OpenAI-compatible multimodal models.

Who it is for — When you use DeepSeek Harness for coding tasks that require processing screenshots or charts, install this plugin. Users who rely solely on text-only models without any visual input needs do not need it, as the plugin focuses on visual analysis enhancements.

Watch out — The build scripts are blocked by pnpm during installation and require manual permission to complete registration. It executes shell commands but this is only a static check prompt. After configuration, select a model from the DeepSeek (SideSight vision) provider to handle attached images.

The verdict — I would install it because it supports OpenAI-compatible multimodal model analysis for visual content, but it requires manual permission for builds to complete installation.

What it is — Provide structured Markdown maps for AI agents to operate specific software.

Who it is for — When Claude Code handles complex Adobe After Effects tasks on Windows, this plugin supplies the required operational knowledge. For agents whose tasks stay within built-in tools or local environments, this plugin is not applicable.

Watch out — Sandbox test passed. Successfully registered in a fresh profile. No obvious issues found.

The verdict — I would install it for Claude Code to automate Adobe After Effects on Windows because it provides verified observation and control methods.

What it is — Integrates as DeepSeek Harness tool bundle, enabling agents to use HTTP API for public plaza posts and replies.

Who it is for — Agents running DeepSeek Harness for collaborative research across hosts can install this to access public plaza tools for insights. Users focused solely on private tool integration without public discovery needs can skip it, as the plugin is designed for serendipitous agent interactions.

Watch out — No obvious issues found. Sandbox testing succeeded in a new profile and was registered into the profile.

The verdict — I would install it if the agent needs public insights from other agents, because it enables serendipitous interactions without signup.

What it is — Logic probe plugin verifies verifiable claims in design documents and refactoring plans for AI coding agents.

Who it is for — When using DeepSeek Harness to review design documents or refactoring plans and check if claims match the codebase, this plugin is suitable. Users who only work with code snippets that do not require design document or plan verification do not need it.

Watch out — Sandbox testing passed on a fresh profile. The plugin executes shell commands and requires user confirmation before running model extractions. No other obvious issues found.

The verdict — I would install it because it generates structured findings with exact file:line evidence for claim verification, which adds direct value to agent sessions involving design plans.

What it is — DeepSeek Harness event audit plugin that observes event types, distribution patterns, counts and recent events to help plugin authors understand internal processes.

Who it is for — When developing a Harness plugin and needing to monitor event types and distribution patterns, install this. Terminal users who do not care about internal event flows can skip installing it.

Watch out — Sandbox testing passed in a fresh profile. Static checks indicate possible shell command execution. Verified to run with settings hot-reload and waterfall event capture in local DSH source environment.

The verdict — I would install it because it provides event auditing capabilities, but only if you are a plugin author debugging internals.

What it is — Observes agent task-model-result, distills routing table and routes to cheapest successful model.

Who it is for — When executing code tasks with deepseek-v4-flash, users needing token-efficient routing are suitable for installation. Users not using deepseek-v4 series models can skip.

Watch out — Sandbox test passed, installed in new profile and registered by harness. Depends on unreleased @deepseek-ai/* packages, requires monorepo build. No obvious pitfalls found.

The verdict — I would install it because it enables token-efficient routing for code tasks with flash model, but only after the dependencies are released to npm.

What it is — DeepSeek Harness request context analysis plugin that records changes in model requests and cache reuse.

Who it is for — When you need to view changes in model request contexts in DeepSeek Harness, this plugin is suitable to install. If you do not use DeepSeek Harness for model request calls, you do not need to install it.

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

The verdict — I would install it because it focuses on request context analysis with zero overhead, but the premise is that you need to debug model calls.

What it is — vpshub is a plugin for DeepSeek Harness that provides SSH ledger to let agents discover, test, execute on, and transfer files to cloud servers.

Who it is for — If you use DeepSeek Harness agents to manage multiple cloud servers requiring SSH connections and file transfers, vpshub fits your needs. If you primarily rely on local model inference without any remote operations, you can skip it.

Watch out — Sandbox test confirms the install command succeeds, but since the package.json lacks dsh.bundle, harness treats it as an ordinary dependency and does not create a profile layer. Installation requires running npm install dsh-vps-hub in the profile directory and editing cordis.patch.yml to insert the configuration. No obvious pitfalls found.

The verdict — I would install it because it allows agents to securely operate remote servers using local key paths and memory-only passwords. However, only if your workflows involve frequent cloud server management, as otherwise the added utility is limited.

What it is — dsh-nocturne-memory connects DeepSeek Harness to Nocturne Memory, offering session-start boot protocol and tools for memory read, search, create and update.

Who it is for — Users who need to load core memories and recent context in DeepSeek Harness plugin system are suitable for installing this plugin. Users who only perform short interactions without requiring external memory server support can skip installing it.

Watch out — Sandbox testing passed: successfully installed and registered in a new profile. Static check indicates it will execute shell commands. It requires your own Nocturne MCP server as the backend, with memories stored only on your server.

The verdict — I would install it because it provides session-start memory loading tools, but only if you have a Nocturne MCP server available.

What it is — DeepSeek Harness plugin providing real browser access via Cloudflare's headless Chrome for JS-rendered pages, screenshots, and PDFs.

Who it is for — When your DeepSeek Harness agent needs to fetch from public webpages requiring JS rendering, this plugin supplies the real browser capability.

If your workflow only involves non-public web content or data processing not requiring browser tools, this plugin is not applicable.

Watch out — Requires Cloudflare API token and account ID configuration. Sandbox testing confirmed successful registration in a fresh profile and integration with Harness. It executes shell commands. No obvious issues found.

The verdict — I'd install it if handling dynamic web content is part of the agent workflow, because it addresses limitations in the built-in web_fetch tool for JS-rendered pages.

What it is — deepseek-vl-support plugin routes image files to OpenAI-compatible vision endpoints to give text-only DeepSeek models vision in supported agents.

Who it is for — When you use a text-only DeepSeek model in DeepSeek Harness and need it to understand pictures in your project files, install this plugin. People who do not need vision capabilities for their models can skip it.

Watch out — Installation requires Node.js 18+ and an account with a vision service. The install command succeeds in tests, but it installs as an ordinary dependency and does not become a profile layer. No obvious issues found.

The verdict — I would install it because it enables vision for DeepSeek models, but only after configuring the vision endpoint as described in the README.

What it is — This is a desktop plugin for DeepSeek Harness that supports native window and tray features.

Who it is for — Windows users who want native desktop interface support when running DeepSeek Harness. Users on non-Windows platforms can continue with the web version because they may not need native tray functionality.

Watch out — Build scripts are blocked during installation, requiring manual approval to complete registration. Sandbox test confirms it can be installed but registration is incomplete. No obvious issues found.

The verdict — I would not install it because even though it provides desktop integration, manual approval of the build is still required after installation, adding extra steps.

What it is — Allow favoring DeepSeek Harness sessions and reopening them from a global cross-Workspace center.

Who it is for — If you manage sessions across multiple workspaces using DeepSeek Harness Web and want a centralized favorites list for quick reopening, this plugin is suitable. If you prefer a single-workspace setup or the built-in session browser without extra features, you can skip installing it.

Watch out — When installing from GitHub source, you need to manually allow the build script execution. Sandbox testing passed with successful registration in a fresh profile. No obvious issues found.

The verdict — I would install it because it centralizes session favorites across workspaces, but only if you are comfortable with the shell command execution requirement.

What it is — AI converts repeated development commands, prompts, and habits into one-click actions for DeepSeek Harness.

Who it is for — Developers repeating flutter run commands in Flutter projects using AI can access them through one-click buttons provided by dsh-dev-actions. For users not relying on DeepSeek Harness AI tools, this plugin's one-click entry is mainly for AI-driven workflows.

Watch out — Sandbox test passed: installed in new profile and registered in harness, action panel displays normally. No obvious issues found. Commands run in managed Shell and session sandbox, AI instructions default to input box, user retains final control.

The verdict — I would install it because it allows AI to actively maintain actions in the dev loop, saving time on manual organization; the premise is using DeepSeek Harness with repeated development commands or prompts that need to be preserved.

What it is — DeepSeek Harness plugin for local snapshots and WebDAV cross-machine sync.

Who it is for — If you switch between multiple machines when using DeepSeek Harness and need to sync session logs and workspace data. If you focus only on single-machine operations with DeepSeek Harness and avoid network data transfer.

Watch out — Tested successfully in a fresh profile and registered into harness. However, an API key or token is required. No other obvious issues detected.

The verdict — I would install it because it enables automatic backup retention and stale archive cleanup, which is useful for ongoing use, provided credentials are configured first.

What it is — This is a DeepSeek Harness settings page that limits each configured model service to the context window it can actually hold.

Who it is for — When configuring multiple model services in DeepSeek Harness, you need to limit each to its actual context window capacity.

If your application only involves a single model, you do not need to manage multiple model context windows.

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

The verdict — I would install it if I need to limit multiple models to their actual context windows, because it modifies settings through the user layer without interfering with official compaction.

What it is — It injects a greeting instruction at the start of every session and remembers the user's name.

Who it is for — When you want the agent to greet you differently every session and remember your name. If you don't want to remember the name, you can skip installing it.

Watch out — Sandbox testing passed, installed in a new profile and registered by harness. No obvious pitfalls found.

The verdict — I would install it because it injects a greeting at session start and remembers the name, which has value when sessions start frequently.

What it is — This plugin connects DeepSeek Harness lifecycle events to the TMCRA Memory API for cross-tool project memory recall and updates.

Who it is for — When conducting multi-conversation projects in DeepSeek Harness, maintaining memory continuity across sessions is necessary. If you operate within a single tool without requiring cross-application memory sharing, installation is unnecessary.

Watch out — Sandbox testing confirmed successful installation and registration in a fresh profile. Static checks indicate shell command execution. No obvious issues were found.

The verdict — I would install it if cross-tool memory continuity is needed, because it enables seamless project continuation across applications without repeated introductions.

What it is — DeepSeek Harness plugin enabling same-runtime cross-session discovery and communication.

Who it is for — When you need to coordinate multiple sessions within one DSH process, it is suitable to install. Those who use only independent runtimes can avoid installing it.

Watch out — It installs in a new profile and gets registered by the harness. No obvious issues found.

The verdict — I would install it because it integrates cross-session capabilities into the existing runtime without additional processes.

What it is — Integrates Jina AI tools into DeepSeek Harness, providing 12 model tools for search, academic retrieval, reading, screenshot, embedding, and more.

Who it is for — Academic researchers using DeepSeek Harness who need arXiv or SSRN paper retrieval tools. Those whose tasks can be completed with the model's internal knowledge may not need to install these external tools.

Watch out — Tested successfully in a fresh profile, registering with Harness. Users in mainland China require a VPN to access Jina domains. No obvious issues found.

The verdict — I would install it because it provides dedicated tools for academic searches returning official direct links.

What it is — harness-desktop provides a desktop client for DeepSeek Harness that can be started immediately without terminal or environment configuration.

Who it is for — When you need a desktop client for DeepSeek Harness without terminal or environment configuration, this plugin is suitable for you. If you mainly run tasks in the command line environment or have already used DeepSeek Harness through the web interface, this desktop client provides a supplementary interaction method.

Watch out — Sandbox test passed: it was installed in a new profile and registered into the profile by harness. The package size is large, about 130-230MB, and currently not signed. No obvious installation issues found.

The verdict — I would install it because it includes agent evolution features such as memory precipitation and skill extraction from similar tasks.