Every DeepSeek Harness plugin

What it is — Switches the engine for DeepSeek Harness sessions based on presets.

Who it is for — Users who need to switch engines in the same session. Users who only want to use DeepSeek engine.

Watch out — Installed from GitHub source. Sandbox test passed in new profile.

The verdict — I would install it because it allows switching engines without restarting sessions or modifying DSH.

What it is — dsh-report-studio converts DeepSeek Harness sessions into work reports with verifiable receipts, including daily, weekly, handoff, and article templates.

Who it is for — When an agent has finished work and needs to hand it over to the next person or generate daily/weekly reports, dsh-report-studio generates verifiable session-to-deliverable reports with deterministic session extraction and verifiable receipt blocks. If session data extraction does not rely on the durable session event log but uses model memory directly, this plugin is not needed.

Watch out — Installation requires Node.js ≥ 18. Source installation requires manually allowing the build script. Sandbox testing passed on a fresh profile with registration into the profile. No obvious issues were found.

The verdict — If the agent needs to convert sessions into deliverable work reports with verifiable receipts, I would install it because it provides deterministic session extraction and safe persistence.

What it is — Exposes update/retrieve/subscribe services for local memory and knowledge in DeepSeek Harness.

Who it is for — Users requiring persistent memory management when using DeepSeek Harness Agent Loop for agent construction. Those without Rust stable and C toolchain.

Watch out — Sandbox testing shows it can be installed but the build script is blocked by pnpm, requiring manual allowance to register. No other obvious issues found.

The verdict — I would install it from source after manually allowing the build script because it provides durable transactional storage with Rust backend.

What it is — DeepSeek Harness plugin for multi-repo workflow isolation.

Who it is for — When working across multiple code repositories and needing to isolate context to prevent pollution, this plugin provides entry agent and controller. If your work is limited to single repository local maintenance without model tiering, this plugin does not fit.

Watch out — Tested through in sandbox: installed in a new profile and registered into the profile. No obvious pitfalls found.

The verdict — I would install it because it maintains hash-chained ledger for state archiving.

What it is — The plugin groups DeepSeek Harness loaded plugins by official vs self-installed source and provides a sidebar Plugin Catalog panel for searching and toggling self-installed plugins.

Who it is for — If you need to quickly distinguish official plugins from self-installed ones and enable or disable specific self-installed plugins via the composer in DeepSeek Harness, this plugin suits your use case. If you only need a simple plugin list or primarily use official plugins, you do not need to install it.

Watch out — Manual approval of the build script is required when installing from source. Sandbox testing passed: successfully registered and enabled in a fresh profile. No obvious installation or runtime issues found.

The verdict — I would not install it because it only adds grouping to the plugin list and does not enhance core agent or token capabilities.

What it is — It displays your DeepSeek account balance in the sidebar footer above Settings, refreshing automatically every 60 seconds.

Who it is for — If you use the DeepSeek Harness web interface and need to view your account balance in the sidebar, this plugin is suitable. If you use command-line tools or do not use the web interface, this plugin is not applicable.

Watch out — It installs successfully as a regular dependency (its package.json lacks dsh.bundle, so it does not become a profile layer). You need to add the plugin entry to cordis.patch.yml to activate it per the README. No obvious issues found.

The verdict — I would install it if I need to display account balance in the DeepSeek Harness sidebar because it supports hot-loading of configuration changes without restarting the web server.

What it is — dsh-tui-app is an Ink-based terminal UI plugin that allows controlling four agents—hermes, claude, codex, and dsh—in a single terminal session.

Who it is for — When you need to command multiple DeepSeek Harness agents from the terminal for coding or multi-task work, this plugin provides a unified interface. If you only use a single agent or do not rely on terminal UI, you do not need it.

Watch out — The npm command installs successfully, but the plugin does not declare dsh.bundle in package.json, so it is installed as an ordinary dependency rather than a profile layer. It executes shell commands. No obvious issues were found.

The verdict — I would install it because it supports commanding multiple agents from one terminal, but it does not become a profile layer after installation and must be configured as described in the README.

What it is — The structured multi-panelist review workflow for DeepSeek Harness.

Who it is for — This is suitable for use in DeepSeek Harness review workflows. Not suitable for users who do not prefer blind scoring, because the workflow is designed with panelists blind to each other.

Watch out — Sandbox test passed in a new profile and was registered by harness. Installation from source requires manually allowing the build script. Static check indicates it will execute shell commands.

The verdict — I would install it because it supports re-reviews.

What it is — Provide a project-agnostic multi-agent collaboration protocol for DeepSeek Harness, including loop guards, six-field handoff format, and risk-tiered review routing.

Who it is for — When you need multi-agent collaboration with unified workflow and delivery sequence, using DeepSeek Harness. If you only involve single-agent tasks or your workflow is already defined by other mechanisms, you do not need to rely on this plugin.

Watch out — Requires manual approval of build script when installing from source. Executes shell commands. Sandbox test passed: registered successfully in a fresh profile.

The verdict — I would install it because it provides a fixed verify→commit→push→review→close sequence and loop guards.

What it is — Adds color-gradient animated whale favicon to DeepSeek Harness web pages and rainbowizes brand elements.

Who it is for — Users who use DeepSeek Harness web and want color-gradient flowing whale favicon are suitable to install. If they mainly use DeepSeek's other interfaces, they do not need this plugin.

Watch out — Requires manually allowing the build script from source install. Sandbox test passed. No obvious issues found.

The verdict — I would install it because it supports two installation methods.

What it is — This plugin supports attaching any file types in the DeepSeek Harness Web UI as pathless @-mentions.

Who it is for — Users who need to reference various file types in messages in the DeepSeek Harness Web UI are suitable for installing this plugin. Users who only handle pure text input without external file references do not need this plugin.

Watch out — No obvious issues were found. The sandbox test passed in a new profile, with the plugin successfully registered by the harness.

The verdict — I would install it because the agent can read the referenced files directly using its own tools instead of guessing their locations.

What it is — It provides a token-efficient persona runtime for DeepSeek Harness that supports local rendering of catgirl style without consuming model tokens.

Who it is for — When performing coding tasks in DeepSeek Harness and wanting to add local catgirl flavor to the output without increasing token costs, this plugin is suitable. It is not suitable for tasks requiring parallel subagents because headless processes do not wait for background subagents to complete.

Watch out — Measured results show 67% reduction in first request new input and 66% reduction in stable cache hits with no quality degradation. Building the client plugin from source has npm registry dependency chain damage issues. Headless one-time processes do not wait for background subagents to complete is a known limitation.

The verdict — Under the premise of no quality degradation, I would install it because it avoids the per-request repeated token costs of traditional persona injection.

What it is — This is a pure client CSS theme plugin for DeepSeek Harness Web providing dual-mode themes for deep sea and shallow sea with plankton animation background and deep diving work state enhancement.

Who it is for — If you are using DeepSeek Harness Web in the chat bar and want to add plankton animation background, you can install this plugin. If you prefer using light mode without caring about deep diving work state enhancement, you can choose other theme plugins.

Watch out — Sandboxes test passed: it was installed in a new profile and registered by harness into the profile. It will execute shell commands.

The verdict — I would install it because it provides zero-intrusion token level coverage and plankton animation, but the premise is that users accept shell command execution during installation.

What it is — DeepSee turns DeepSeek Harness into a multimodal coding workspace with vision and smart model routing.

Who it is for — Developers using DeepSeek V4 Flash or Pro for UI coding who need visual verification and cost-based model routing will benefit from it. Users whose work consists solely of text processing without visual interfaces can use standard text-only DeepSeek sessions.

Watch out — Sandbox testing confirmed successful installation and registration in a new profile. No obvious issues found.

The verdict — I would install it because it enables Gemini to perform visual checks at milestones and before delivery, returning PASS or defect locations.

What it is — dsh-vision plugin lets text-only DeepSeek models support image viewing, pasting, OCR, and visual Q&A through the view_image tool.

Who it is for — When developers need to handle image attachments in DeepSeek model conversations, this plugin can be installed. Those using models that natively support images do not need it.

Watch out — No obvious issues found. Sandbox testing passed successfully. Requires configuration of API keys for VLM providers.

The verdict — I would install it because it enables reliable image handling without session crashes in text-only setups.

What it is — Provides trajectory anchor badges and click-to-rollback for DSH session rollback.

Who it is for — Developers debugging multi-turn conversations using DeepSeek Harness who need visual rollback points are suitable for installing this plugin. Users who do not rely on DSH built-in fork/branch features can skip it, because DSH provides built-in fork/branch functionality.

Watch out — Sandbox test passed. The plugin was registered into the profile. No obvious issues found. The repository is scrapped.

The verdict — I would not install it because the development direction has been overturned.

What it is — Transform DeepSeek Harness into a desktop-style workbench with sidebar file tree, editing, preview, and terminal.

Who it is for — If you are working on multiple projects in DeepSeek Harness and need a left sidebar for projects and files, middle for editing and preview, and right for AI chat, this plugin provides the corresponding desktop-style transformation. If your task only involves running specific commands without needing file management or AI dialogue integration, then you can skip this plugin.

Watch out — The sandbox test passed: the plugin is installed and registered into the profile in a fresh profile. No obvious issues found.

The verdict — I would install it because the sandbox test passed and it uses the Cordis configuration layer to replace the layout without modifying the DSH installation directory.

What it is — iterate-plugin is a DeepSeek Harness plugin that integrates the iterate skill into autonomous closed-loop code iteration, supporting multi-round parallel review and deterministic deduplication convergence.

Who it is for — When using DeepSeek Harness for code iteration that requires AI to complete multi-round review, fix, and self-stopping loops, iterate-plugin is suitable to install. If you only need one-time code modifications or do not involve autonomous iteration skills, you can choose not to install it.

Watch out — It ran successfully after installation and registration by Harness in a new profile. It can execute shell commands. No other obvious installation or runtime issues were found.

The verdict — I will install it because it provides a build-free Web UI layer and specific iteration tools integrated into Harness.

What it is — DeepSeek Harness client plugin manager that provides grouped inventory and marketplace browsing for web settings.

Who it is for — If you are using DeepSeek Harness in web mode and need to manage client plugin lists and quickly browse install commands, you can install it. If you focus more on server-side plugin management, it does not provide related functionality.

Watch out — Sandbox test passed: installed in fresh profile and registered by harness. It will execute shell commands, though static check is only a prompt. No obvious issues found.

The verdict — I would not install it because it is open source but has no declared license.

What it is — Tools inside dsh agents to search npm and awesome-dsh-plugin lists for dsh plugin details.

Who it is for — When building dsh agents that need to find plugins for tasks like GitHub release monitoring, dsh_search_plugins can get results directly from npm and lists. If your workflow does not rely on agent internal tool calls, this plugin can be replaced to simplify dependencies.

Watch out — Sandbox test shows it can be installed but the build script is blocked by pnpm, requiring manual allowance of the build script from source install to complete registration. No other obvious issues found.

The verdict — I would install it because it enables direct plugin discovery inside agents, but the build script requires manual approval to register fully.

What it is — This is a vision plugin for DeepSeek Harness that converts images to structured JSON evidence for text-only models to see.

Who it is for — When DeepSeek Harness reads images, this plugin suits letting text models access images and output structured data. If you only configure a single engine, you do not need the multi-engine selection feature of this plugin.

Watch out — Sandbox test passed: successfully installed in a new profile and registered by Harness. It will execute shell commands. Engine configuration is required for normal operation.

The verdict — I would install it because it supports multi-engine selection, but the premise is that the user needs to configure multiple providers.