Introduction

DeepSeek Harness (DSH) is a framework focused on agent development, with its core strength in extending functionality through plugins. When handling video generation tasks, developers usually need to write code to orchestrate the processing flows of scripts, prompts, images, and audio. DeepSeek Harness Video-Director turns this process into visual node-based operations, aiming to lower the barrier to developing video generation workflows.

What It Is

DeepSeek Harness Video-Director is a visual, node-based video director plugin, maintained by developer chiphoton and released under the MIT license. Built on top of DeepSeek Harness, it transforms script, image, audio, and video generation into a connectable canvas, providing out-of-the-box workflows.

Core Capabilities

The plugin provides the following core features:

  1. Visual production graph: Connect elements such as text, images, audio, video, workflows, previews, and saved outputs on an infinite canvas.
  2. Local-first and multi-provider: Supports Ollama, ComfyUI, OpenAI-compatible endpoints, and Codex Plan. By default, it supports local models such as Qwen3.8-27B and MiniMax-H3.
  3. Project awareness: Each video project has its own canvas, conversation sessions, task queue, and immutable resources.
  4. Extensible design: Supports importing reviewed API-format ComfyUI workflows or packaging them as declarative vd-node packages.

Installation and Activation

Before installing, ensure your environment meets the following requirements: Git, Node.js ^22.19.0 or >=24, pnpm 11.7.0, and the DSH CLI installed. Although NVIDIA DGX Spark is the recommended local deployment environment, it is not required.

Run the following steps to install the plugin into the DSH web profile:

git clone --branch main --single-branch https://github.com/chiphoton/DeepSeek-Harness-Video-Director.git
cd DeepSeek-Harness-Video-Director
pnpm install --frozen-lockfile
pnpm run build
dsh plugin --profile web add "file:$PWD"
dsh web

After installation, simply run dsh web to start it.

Typical Usage

Building a workflow typically includes the following steps:

  1. Add nodes: Double-click on an empty canvas to add vd-nodes. For example, add a text node, a prompt-enhancement node, and an H3 video node.
  2. Connect nodes: Connect the nodes in logical order. An example connection path is: Text → Prompt Enhancer → H3 Video → Preview → Save Output.
  3. Configure provider and model: On executable nodes, set the provider (such as Ollama or ComfyUI), select the corresponding model or registered ComfyUI workflow, and enter the generation prompt.
  4. Run and save: Click Run to execute the workflow. Use the save function in the top bar to persist canvas edits.

Caveats and Use Cases

The plugin runs with the permissions of the current dsh process. Before installation, review the source code and license. The plugin does not silently install models, ComfyUI custom node packages, or external services; users are responsible for managing underlying model dependencies.

Conclusion

DeepSeek Harness Video-Director is suitable for developers who need to visualize video generation pipelines. It simplifies multi-model, multi-step video generation orchestration through a node-based canvas. For more details, refer to the community directory or the GitHub repository.