dsh-generation-image
Run the following command in DeepSeek Harness:
dsh plugin install sfyyy/dsh-generation-image
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install sfyyy/dsh-generation-image in your terminal to install this plugin; the source code is available at https://github.com/sfyyy/dsh-generation-image .
About this plugin
DeepSeek Harness agents run on a text-only model; the moment a task asks the agent to draw something or restyle an upload, there is no execution path. dsh-generation-image fills that gap: it registers a generate_image tool inside the agent session so the agent can call your own OpenAI-compatible image endpoint mid-reasoning and drop the resulting image into the conversation as a first-class message.
The plugin covers both text-to-image and image-to-image. For text-to-image the agent simply passes a prompt; for image-to-image it extracts attachment IDs from the current conversation and routes the call to the /images/edits endpoint for modification, restyling, or multi-image composition. Generated images are durably saved as DSH attachments and rendered as clickable thumbnails in the Web UI with enlarge, download, and multi-image navigation. In parallel, image blocks are rewritten to text markers before any request reaches the text model, keeping inference safe while the session log and UI retain real pixels.
Built for developers who already run DeepSeek Harness agent workflows and have access to any OpenAI-compatible image API such as OpenAI, vLLM, local gateways, or third-party relays. No extra server to deploy: fill in a baseUrl and an API key and you are live. Also a natural fit for design teams that want to close the describe, generate, refine, finalize loop entirely inside one chat session.
Use Cases
- Agent needs to generate an illustration or cover image on the fly during a chat
- User uploads multiple images and asks the agent to restyle, edit, or compose them
- Closing the describe-to-final-image loop entirely inside one agent session
Best For
- AI developers running DeepSeek Harness who need image capabilities in their agent loop
- AI application teams with access to any OpenAI-compatible image endpoint such as vLLM or local gateways
- Design and creative teams who want to fold image generation into a conversational workflow
Related Plugins
ModLens is a vision plugin for DeepSeek Harness that gives text-only models sight by reading images pasted directly into chat, with zero-config setup and multiple vision engines.
On-demand vision for text-only DeepSeek Harness agents: built-in free keyless vision chain and 14 vision tools, routing image turns as tool calls to vision models with pixel fidelity, no Python needed, one-command install.
Give text-only models in DeepSeek Harness eyes, enabling image Q&A, long-screenshot OCR, UI restoration, and GUI visual tasks.
An armor-breaking plugin for DeepSeek: rewrites instruction priority, expands refusal-blocked phrases, persists across sessions, and shows a green active indicator.