Every DeepSeek Harness plugin

What it is — Unifies DeepSeek Harness extensibility seams into a declarative DSL with a runtime fabric service, fabric_extend and fabric_inspect tools, and a capability-graph settings page.

Who it is for — DSH plugin developers can register code-based extensions via the fabric service. Model users who do not need to register pure declarative extensions through tools do not need to install this plugin.

Watch out — Installation from source requires manually permitting the build script to execute. Sandbox test passed in a fresh profile, successfully registered into the profile. No obvious runtime issues found.

The verdict — I would install it because it makes runtime extensions a programmable primitive.

What it is — Converts articles from DeepSeek Harness sessions into editable Excalidraw canvases.

Who it is for — When processing articles in DeepSeek Harness conversations using the Agent, if you need structured visualization, you can install dsh-diagram. If you only need generation of static images, you do not need to install this plugin.

Watch out — Installation requires Node.js ^22.19.0 or >=24.0.0 and pnpm >=10. It will execute shell commands. Sandbox testing passed: installed in a fresh profile and registered into the profile by harness. No obvious issues found.

The verdict — I would install it because it enables persistent editable Excalidraw canvases and autosave within the DSH session, which adds value for article analysis tasks, but only if the environment has DSH Web bound to 127.0.0.1.

What it is — It integrates text, vision, and image generation APIs into a Mix model with automatic routing.

Who it is for — Users needing their agent to analyze user-uploaded images and screenshots would install this. Those only performing text reasoning without image handling do not need it.

Watch out — Sandbox test passed: installation in a fresh profile succeeded and it was registered to harness. No obvious issues found.

The verdict — I would install it because it supports image generation and editing with session-scoped history.

What it is — dsh-lab provides a verified list of DeepSeek Harness plugins that passed installation and static scan tests.

Who it is for — When you are building a profile that includes multiple DeepSeek Harness plugins, dsh-lab can help you find ones that have been tested to install successfully. If you only want to quickly try out a few plugins without caring about their installation compatibility, dsh-lab's verification results are not needed.

Watch out — The installation test passed in a new profile and the plugin was registered by harness. Static scan found no issues.

The verdict — I would install it because it documents six vanished repositories from community directories and four reproducible install frictions, which helps in maintaining plugin compatibility.

What it is — dsh-ponytail registers 6 ponytail skills to DeepSeek Harness.

Who it is for — Suitable when using DeepSeek Harness model to code and wanting the model to automatically adopt a lazy but minimal code approach. If you do not use DeepSeek Harness, you need not install this plugin.

Watch out — Sandbox test passes: it installs in a new profile and registers with harness. No obvious pitfalls found.

The verdict — I would install it because it has zero runtime dependency.

What it is — Plugin dsh-ecc ports 273 skills from affaan-m/ECC to DeepSeek Harness.

Who it is for — When performing engineering methodology work on DeepSeek Harness, users missing agentic-engineering skills can install dsh-ecc. Users not involved in pattern and data development can skip this plugin.

Watch out — Sandbox test passed: installed into a fresh profile and registered by harness. No obvious issues found. Static checks showed no issues.

The verdict — I would install it because 95.8% of the skills have been ported, with the remaining depending on ECC-specific infrastructure.

What it is — llm-adaptive adds the adaptive provider to DeepSeek Harness, enabling automatic model routing based on per-request complexity classification.

Who it is for — When using DeepSeek Harness to handle multi-turn LLM requests that require dynamic routing based on request complexity, this plugin provides per-request complexity classification for automatic provider selection. Users who do not configure the classifier section of their pool.json will not be able to utilize the adaptive routing feature.

Watch out — Sandbox testing passed: the plugin was installed in a new profile and registered into the harness profile. No obvious pits found. It will execute shell commands but no issues were observed during testing.

The verdict — If I were setting up a DeepSeek Harness project needing adaptive routing for requests of varying complexity, I would install it because the fail-open mechanism ensures requests are never blocked.

What it is — This plugin integrates a GrokBot-style animated avatar featuring synchronized typewriter status effects in the DeepSeek Harness web interface.

Who it is for — Users benefit from this plugin when observing dynamic avatar feedback in the status indicator during model deep diving in DeepSeek Harness web interface. Users who do not use the web version of DeepSeek Harness do not need this plugin.

Watch out — Sandbox test passed with successful registration in a new profile. No obvious issues found. The plugin is pure client-side with zero runtime dependencies.

The verdict — This plugin is worth installing because it adds lively dynamic effects to the thinking interface and runs stably.

What it is — A DeepSeek Harness plugin providing the wuyun-liuqi skill pack.

Who it is for — Users studying wuyun-liuqi theory using DeepSeek Harness are suitable for installing this plugin. Users who do not use DeepSeek Harness can skip it.

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

The verdict — I would install it because it registered 36 skills.

What it is — dsh-taskboard is a task board plugin for DeepSeek Harness that supports creating tasks with project and model assignment and manual or cron execution.

Who it is for — It suits users who collaborate on projects and assign specific models for task execution. Users without project collaboration or cross-project task needs do not need to install this plugin.

Watch out — Sandbox test passed and it registers in a fresh profile. Static check indicates execution of shell commands. No obvious pitfalls found.

The verdict — I would install it because it enables structured project task collaboration and automatic todo to in-review flow.

What it is — DeepSeek Harness plugin providing tavern management panel for multiple character cards, multiple worldbooks, and switchable presets.

Who it is for — Users managing multiple character cards and worldbooks during DeepSeek Harness role-playing are suitable for installing this plugin. Users not requiring multiple character cards or worldbooks do not need to install it.

Watch out — Sandbox testing passed with successful registration in a new profile. No obvious issues found.

The verdict — I would install this plugin if managing multiple character cards and worldbooks is required, as it allows one-click generation of agent.cordis.yml.

What it is — Integrates interactive SSH terminal, supporting driving connected servers from the main conversation in DeepSeek Harness.

Who it is for — When you need to drive a connected server via SSH commands in DeepSeek Harness conversation, this plugin fits. If you prefer managing servers directly in a local terminal without dialogue tools, this plugin does not apply.

Watch out — Sandbox test shows installation is possible but pnpm blocks build scripts including cpu-features@0.0.10 and ssh2@1.17.0, requiring manual approval for full registration. No other issues found.

The verdict — I would install it for conversation-based SSH terminal control, but only after manually allowing the build script.

What it is — In DeepSeek Harness Web, when chat messages contain Mermaid fenced code blocks, local rendering previews Mermaid diagrams as images, supporting external rendering servers.

Who it is for — In DeepSeek Harness Web chat scenarios, when messages contain Mermaid fenced code blocks, this plugin can preview Mermaid code blocks as images. If you prefer external rendering servers over local rendering or your chat content does not involve Mermaid syntax, you can skip installing this plugin because it focuses on specific rendering mount points.

Watch out — Sandbox test passed in new profile; the plugin was registered into the profile. No obvious issues found.

The verdict — I would install it because it defaults to local rendering and renders Mermaid diagrams as thumbnails, supporting click to toggle size in place without popping up a new window.

What it is — The plugin adds an "Allow for this task" button to DSH Web approval cards, automatically allowing matching danger-full-access escalations only for the current task.

Who it is for — When using an AI agent in DSH Web to handle high-risk tasks requiring danger-full-access escalation and wanting to reduce repeated approvals in the same task, this plugin is suitable. Users running DSH in headless mode without UI interface will not find this plugin applicable, as it only extends the Web approval cards.

Watch out — No issues were found in static checks. The sandbox test passed with successful registration in a fresh profile. No obvious problems detected.

The verdict — I would install it because it scopes the authorization to expire after the current task, but only if using DSH Web and the agent requests such escalations.

What it is — dsh-forge is a runtime extension suite for DeepSeek Harness, providing plugin market, agent teams, and cross-session mailbox.

Who it is for — When you use DeepSeek series models to perform agent team tasks and need cross-session mailbox and skill management, it is suitable to install it. If you use non-DeepSeek series models, you can choose other task routing schemes.

Watch out — Tested successfully in a new profile and registered by harness. After install, need to restart DSH. Shell commands are executed for reviewed plugins only.

The verdict — I would install it because it provides task-aware routing and skill manager, but only if you accept restart after install.

What it is — Supplements models.dev reasoning levels for unconfigured third-party DeepSeek Harness models.

Who it is for — Users who lack reasoningEfforts configuration when using llm-pi-ai agent models can install this plugin. Users who have already configured reasoningEfforts do not need 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 can supplement reasoningEfforts for unconfigured models in llm-pi-ai and no build scripts are required.

What it is — Automatic TypeScript type checking plugin for DeepSeek Harness.

Who it is for — When a model is editing or creating TypeScript files, this plugin is suitable. It is not needed if the project does not contain TypeScript files.

Watch out — Source installation requires manually allowing its build script. Sandbox test shows it can be installed but the build is blocked and manual approval is needed to complete registration. No obvious issues found.

The verdict — I would install it because it automatically catches type errors in TypeScript files, but installation requires manually allowing the build script first.

What it is — Enhances the DeepSeek Harness model selector with searchable filtering by provider, model name, and model ID.

Who it is for — When using the DeepSeek Harness model selector to locate models by provider, name, or ID. If you use a server-side model management system, this plugin is not applicable.

Watch out — Sandbox testing passed: successfully installed in a new profile and registered to the profile. No obvious pitfalls found.

The verdict — I would install it because it provides precise filtering by model ID and passed sandbox tests.

What it is — This is a native HTTP service for running Stata through DeepSeek Harness. It provides macOS and Windows binaries that let AI agents execute Stata commands directly without local Node.js or Python runtimes.

Who it is for — If you use DeepSeek Harness to build an AI agent that needs to run Stata regressions, do-files or econometrics workflows and do not want to install Node.js or Python on the user side, install this plugin.
If you only need to run AI tasks with other programming languages or pure text processing, this plugin is not needed.

Watch out — Installation requires first installing Node.js and DeepSeek Harness globally, then running the dsh plugin command to add the package to the web profile. macOS support is limited to Apple Silicon. Sandbox testing succeeded when installed in a fresh profile and registered with Harness; execution passes.
No obvious issues found.

The verdict — I would install it because it provides a native Stata execution interface that integrates directly into AI agent workflows without additional user-side dependencies.

What it is — dsh-turn-rail is a DeepSeek Harness Web plugin that provides an auto-hiding right-edge turn rail supporting edge-peek show/hide, per-turn tooltip, in-rail search and 30-turn scrolling window.

Who it is for — When using DeepSeek Harness Web to handle sessions, users needing quick previews and searches of turns are suitable for installing this plugin. Users not using DeepSeek Harness Web can skip it because it is a browser interface plugin.

Watch out — Sandbox testing succeeded in registering it in a new profile with harness. It executes shell commands. No obvious pitfalls found. From source install, manually allow its build script first.

The verdict — I would install it because it provides search and scrolling for extended sessions and is a pure browser UI plugin without model involvement.

What it is — Real-time display of session token usage and estimated cost in the composer stats row.

Who it is for — When chatting with a model in the DeepSeek Harness Web GUI and needing to monitor token usage in real time to manage session costs. If using local DSH mode instead of Web GUI, this plugin is not required.

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

The verdict — I would install it because it provides dynamic estimates from the official pricing page, but only if you accept the cost display in the stats line.