xiaozhe7772222/dsh-draw-router

DeepSeek Harness (DSH) 通用图像生成路由器:可从任何兼容 OpenAI 的端点自动发现图像模型。支持 SenseNova、StepFun、Agnes、Qwen、Flux、SD、Imagen 等。| DeepSeek Harness 统一绘图路由器

Project Overview项目介绍

This repository is a native plugin built exclusively for DeepSeek Harness (DSH), designed to add native text-to-image generation capability that the core DSH framework lacks out of the box. After installation, it registers two agent tools, draw_image and draw_list_sources, directly into the DSH agent's tool set, so the agent can generate images on demand just like it runs other commands during a conversation. To get started, users only need to input a base URL and API key for any OpenAI-compatible image generation endpoint, and the plugin will automatically detect available image models on that endpoint via pattern matching against model names, no hardcoding of model lists required.

The plugin supports image generation endpoints from over 30 different providers, including SenseNova, StepFun, Agnes, Qwen, Tongyi Wanxiang, Hunyuan, Seedream, GPT Image, Flux, Stable Diffusion, and Imagen, so it works with most popular commercial and open source image models available today. For providers that do not expose their image models via the standard /v1/models endpoint, users can manually add models via the plugin's built-in REST API. All configuration changes are saved persistently to a local draw-config.json file, so your sources will remain available even after you restart DSH. The REST API also lets users remove existing sources or re-probe models at any time during runtime, giving users full control over their image generation setup.

The plugin requires Node.js version 18 or higher to run, and it is released under the permissive MIT license, which allows free use, modification, and redistribution. If you run into installation issues such as the plugin not being found on startup, you can fix this by creating a correct symlink from your DSH profile folder to the plugin directory, and verifying the plugin is added to the bundles array in your DSH profile package.json. Common runtime issues, such as missing detected models, quota errors, invalid parameters, or authentication failures, are all covered in the detailed troubleshooting section of the README, with step-by-step commands to resolve each issue.

dsh-draw-router 是专为 DeepSeek Harness (DSH) 开发的原生插件,用于弥补 DSH 原生不支持图像生成的短板。它为 DSH 代理注册了绘图工具,让用户可以在对话中直接要求代理生成图片、流程图等可视化内容。用户只需添加任意 OpenAI 兼容接口的 baseURL 和 API 密钥,插件就会自动探测接口下的图像生成模型,无需手动硬编码模型名称。

插件支持几乎所有主流厂商的文生图接口,包括商汤日日新、阶跃星辰、通义万相、混元、豆包即梦、GPT Image、Flux、Stable Diffusion 等。对于不通过 /v1/models 暴露模型的厂商,支持手动添加模型,配置会持久化保存在本地,重启后不会丢失。插件还提供 REST API 支持运行时增删改探测绘图源,满足自定义配置需求。

插件遵循 MIT 许可证,可自由使用修改分发,要求 Node.js 版本不低于 18。首次安装后如果遇到工具不显示的问题,需要检查 cordis.patch.yml 的注入配置,重启 DSH 即可解决。生图过程中遇到的错误大多和厂商配置、账户余额、参数格式有关,可以按照 README 中的故障排查指南逐一解决。

Pre-install check安装前体检Compatibility · Security兼容性 · 安全性 1 warning1 项注意
  • Only 4 stars - very few users, little community feedback星标只有 4,几乎没人在用,遇到问题缺少社区反馈
DSH walks through these 9 checksDSH 会逐条核对这 9 项

Compatibility兼容性

  • DSH, Node, OS and profile requirementsDSH 版本 / Node 版本 / 操作系统 / profile 是否满足要求
  • External dependencies and runtimes (Electron / Python / Docker, ...)外部依赖与运行时(Electron / Python / Docker 等)是否齐备
  • Conflicts with installed plugins: command names, skill / tool names, ports, duplicate MCP registration与已装插件是否冲突:命令名、skill / tool 重名、端口占用、重复 MCP 注册

Security安全性

  • Repo matches the facts registered here; archived or abandoned?仓库是否与页面登记一致,是否归档或长期停更
  • Safety of preinstall / install / postinstall and install.sh / setup.ps1preinstall / install / postinstall 与 install.sh、setup.ps1 是否安全
  • curl|bash, download-then-execute, obfuscation, unrelated domains → stop immediatelycurl|bash、下载即执行、混淆代码、无关域名 → 立刻停止
  • Typosquatting or unmaintained packages among the new dependencies新增依赖里有没有 typosquatting 或无人维护的包
  • Requested permissions vs. what the feature actually needs申请了哪些权限、是否超出功能所需(filesystem / network / shell / clipboard)
  • Any sudo / admin requirement, plus uninstall and rollback是否要求 sudo / 管理员权限,以及卸载与回滚方式

Anything uncertain must be marked unknown with a note on how to confirm it. This site's signal screen is a static snapshot, not a security audit.拿不准的必须标「未知」并说明要我怎么确认。本站的信号筛查是静态快照,不能替代安全审计。

Or use CLI install (for developers)或使用命令行安装(适合开发者)

CLI Install命令行安装

dsh plugin add dsh-draw-router --profile web

把 xiaozhe7772222/dsh-draw-router 加入你的 DSH 配置(web profile)即可启用。

READMEREADME

dsh-draw-router

DeepSeek Harness 缺少生图能力?这个插件让它能。

DeepSeek Harness (DSH) can't generate images? This plugin fixes that.

🐧 Designed & built by 小哲 (@xiaozhe7772222)

Author: 小哲 (xiaozhe7772222) 🐧 · Maintainer: 小哲 (xiaozhe7772222) 🐧

License: MIT DSH Node Author Made with ❤️

填一个 baseURL + API Key → 自动识别绘图模型 → 对话中直接生图

Add a baseURL + API Key → auto-detect image models → draw from chat


English | 中文


English

The Problem

DeepSeek Harness (DSH) is a powerful LLM agent framework, but it has no built-in image generation capability. You can't ask the agent to draw a picture, create a diagram, or generate a visual — it simply doesn't know how.

dsh-draw-router solves this by acting as a universal image generation router. It registers draw_image and draw_list_sources as agent tools, so the DSH agent can generate images just like it runs bash commands — naturally, in conversation.

✨ Features

Feature Description
🎨 Agent-native image gen DSH agent can draw images on demand — just ask "画一张图"
🔍 Auto model discovery Enter any OpenAI-compatible baseURL + API key → plugin calls GET /v1/models and auto-detects image models (never hardcodes model names)
🏭 Multi-provider SenseNova · StepFun · Agnes · Qwen · Tongyi Wanxiang · Jimeng/Seedream · Hunyuan · GPT Image · Flux · Stable Diffusion · Imagen · and any OpenAI-compatible endpoint
🎚️ Manual model add For providers not exposing image models via /v1/models, add models manually
🔌 REST API Add/remove/probe draw sources at runtime; draw via POST /dsh-draw-router/draw
💾 Persistent config Sources persist to draw-config.json, survive restarts
🔒 Key masking API responses never expose full API keys

Showing the opening section of the README — the full document lives in the repository以上为 README 开头摘要,完整文档在仓库内 · View the full README on GitHub →在 GitHub 查看完整 README →

← 上一个 Prev dsh-hub 下一个 Next dsh-git-plugins →