dsh-ramify: A Creative Branching Canvas Plugin for DeepSeek Harness
yanglongyun/dsh-ramify
A creative branching canvas for DeepSeek Harness to generate, compare, and iterate multiple interactive ideas.
Organize, compare, and iterate on multiple creative directions within DeepSeek Harness without leaving your current session.
dsh plugin --profile web add github:yanglongyun/dsh-ramify- Category
- Workflow & Automation
- Platform
- DSH-Plugin
- Author
- yanglongyun
- Distribution
- Plugin
dsh-ramify Key Features
dsh-ramify Repository
dsh-ramify Repo Summary
dsh-ramify is a creative branching canvas plugin (DSH plugin) for DeepSeek Harness, developed by yanglongyun under the MIT license, written in TypeScript. It addresses the challenge of organizing, comparing, and iterating on multiple creative directions generated by an agent: Ramify organizes them into a real-time updating creative tree, allowing you to view, compare, and fork directly within DSH without copying local addresses or reconfiguring models. Core capabilities include native DSH plugin integration, visual branching, multi-format artifact preview, SQLite local persistence, and a set of Ramify tools for the model.
What core features does the dsh-ramify DSH plugin support?
As a native DSH plugin, Ramify uses standard plugin manifests, Cordis services, and Client UI slots without modifying Harness source code. UI submissions are sent to the current model via DSH's official session input capability, so no API key configuration is needed in Ramify. It supports forking from any completed node while keeping old versions intact; HTML, Markdown, SVG, images, videos, and audio artifacts can be previewed directly; data is persisted in SQLite with lightweight frontend polling for changes. On uninstall, the plugin cleans up the runtime it started, and existing Ramify instances are reused rather than shut down.
How to install dsh-ramify?
Installation requires Node.js 22.19 or higher and DeepSeek Harness 0.1.0-rc.6 (current test version) or compatible. The recommended method is from source:
git clone https://github.com/yanglongyun/dsh-ramify.git
cd dsh-ramify
npm install
npm run build
dsh plugin --profile web add "$PWD"
dsh web --port 3099
After startup, click Ramify at the bottom of the DSH left sidebar; the plugin automatically starts the runtime and embeds the workspace into the DSH overlay.
How to use dsh-ramify?
You can start directly from the Ramify input box, or ask the agent in DSH conversation to use Ramify. The plugin registers tools such as ramify_start, ramify_project_create, ramify_node_add, and ramify_node_complete for starting the canvas, creating projects, adding nodes, and writing artifacts. For example, you can say: "Use Ramify to explore three distinctly different landing page directions for this AI search product, and make previewable pages for me to compare."
What is the architecture and data storage of dsh-ramify?
The Ramify UI runs as a DSH overlay iframe, communicating with the current session and model via the DSH Client UI session bridge; the model writes projects, nodes, and artifacts through Ramify tools, and the local Ramify runtime stores them in SQLite and local artifact files. The DSH Web UI runs by default at http://127.0.0.1:3099, and the Ramify runtime at http://127.0.0.1:9519. Data is saved by default in the OS application data directory (macOS: ~/Library/Application Support/Ramify/, Windows: %APPDATA%/Ramify/, Linux: ${XDG_DATA_HOME:-~/.local/share}/ramify/) and persists across upgrades or restarts.
Which DeepSeek Harness versions and platforms are compatible with dsh-ramify?
Ramify requires Node.js 22.19 or higher and DeepSeek Harness 0.1.0-rc.6 (current test version) or compatible. It supports macOS, Windows, and Linux. The plugin does not read, receive, or store model API keys; model calls are handled by the current DSH session.