dsh-imagegen：在 DSH 对话里用 generate_image 生图与改图
dsh-imagegen（Lujinan991）是 DeepSeek Harness 原生 Cordis 插件，非 MCP 包装。智能体通过 generate_image 完成文生图、参考改图与 PNG 蒙版局部编辑，结果写入 DSH 持久附件并在对话预览。支持 OpenAI 兼容 API、异步 task_id 轮询、adaptive 尺寸与 fallbackEndpoints。安装地址 git+https://github.com/Lujinan991/dsh-imagegen.git；兼容 DSH >=0.2.0-rc.2 <0.3.0。配置保存在 cordis.patch.yml，保存后下次调用即生效。
Read Moredsh-agnes-image：在 DSH 会话里生成并落盘 Agnes Image 2.5 Flash 图片
dsh-agnes-image 是 Lijianan-123 维护的 DeepSeek Harness 社区插件，对接 agnes-image-2.5-flash，以单一工具支持文生图、图生图与多图合成，并将 inline base64 结果写入会话工作区 PNG。需 Node.js 22+，安装命令为 dsh plugin --profile web add github:Lijianan-123/dsh-agnes-image，API Key 支持环境变量、凭据存储或密钥文件且每次调用热解析。实现侧重 IPv4 连接稳定、连接级有限重试与单跳取图，零运行时 npm 依赖。
Read More手绘涂鸦风：用 Agent Skill 稳定产出 Sketchnote 叙事插画
SkillHub 社区 Skill「手绘涂鸦风」（@user_0aeadd18/hand-drawn-doodle，1.0.0）面向团建回顾、旅行手账、会议纪要等场景，约束黑色细线、2～3 种点缀色、黑豆人与手写中文标注的 doodle 风格。文章说明其风格 DNA、触发条件、skillhub 安装命令（需 --dir 指向 Agent skills 目录）、image_gen 默认参数，以及官方 prompt 骨架与示例来源，并标明不适用写实/3D/二次元等需求。
Read More产灵 Skill：一个 API Key，让 Agent 调用 gpt-image-2 与 Nano Banana 生图
SkillHub 社区技能 @user_91ce4cec/chanling-skill（v1.0.2）将产灵 API（xdhdancer.top）封装为 markdown Skill：统一 secrets 名 xdhdancer_api_key，总入口路由至 chanling-image，走 POST/GET /v1/videos 异步出图，支持 gpt-image-2 与 Nano Banana 系列。文章说明 skillhub 安装命令、各客户端 skills 目录、令牌配置与自然语言/ curl 典型用法，并提示平台绑定与计费注意。
Read Moredsh-image-bridge：给 DeepSeek Harness 接入外部生图服务
`dsh-image-bridge` 是一个为 DeepSeek Harness (DSH) 开发的独立插件，旨在为不具备绘图能力的模型挂载 `image_generate` 和 `image_edit` 工具。通过支持 OpenAI 兼容接口，用户可为模型配置外部绘图服务。该插件具备图片持久化存储、多服务商支持、备用模型自动切换以及严格的安全管理（如 SSRF 拦截）等功能。用户需通过 npm 安装并重启 DSH，在 Web 界面配置服务商与模型后，即可在对话中直接调用生图能力。
Read Morebanana：在 Agent 里调用 Nano Banana Pro 做文生图与图像编辑
banana（@user_68c7e960/banana）是 SkillHub 社区 design-media 类 Skill，封装 Nano Banana Pro（Gemini 3 Pro Image）的文生图与 --input-image 编辑能力，支持 1K/2K/4K。通过 skillhub install 安装到 Cursor 等 skills 目录后，Agent 按 SKILL.md 用 uv run 调用 generate_image.py，需配置 GEMINI_API_KEY。文章说明安装路径、典型命令与草稿迭代注意点。
Read More智绘图片香蕉：在 Agent 里用 Gemini Nano Banana 做文生图与修图
智绘图片香蕉（@ixhlink/ixhlink-skills-image-banana，v1.0.4）是星华智联科技在 SkillHub design-media 分类发布的 Agent Skill，基于 Google Nano Banana（Gemini 原生图像模型），提供文生图、图生图与对话式修图，侧重画面内文字、4K 与多参考图一致性。安装需 skillhub CLI，并用 --dir 指向 Cursor（~/.cursor/skills/）等客户端 skills 目录。用户提到 Gemini 出图、香蕉模型、4K 或多参考图时可触发。标签为付费类，元数据不要求 API Key 字段。
Read Moreechoflow-banana：在 Agent 里用 EchoFlow 调用 Nano Banana Pro 做图像生成与编辑
社区 Skill @user_9feec992/echoflow-banana（v1.0.0）面向 Cursor 等 Agent：通过清云 EchoFlow API 使用 Nano Banana Pro / Gemini 3 Pro Image，支持文生图、单图编辑与最多 14 张多图合成，分辨率 1K/2K/4K。安装需 skillhub 与 --dir 指向 skills 目录，运行依赖 uv 与 ECHOFLOW_API_KEY，捆绑 generate_image.py 经 chat/completions 落盘 PNG。
Read Moreimage generate：按提示词自动选型 dlazy CLI 出图
image generate（@user_f95f21b8/dlazy-image-generate，v1.0.5）是 SkillHub 社区 design-media 类 Skill，通过 SKILL.md 指导 Agent 调用 dlazy CLI 出图，并根据提示词自动选择较合适的图像模型。标签 requires_api_key=true，安装需 skillhub CLI 并用 --dir 指向 Cursor 等客户端的 skills 目录。本文依据 SkillHub 公开简介与安装文档说明安装路径与用法边界，不编造未公开的 CLI 示例与配置字段。
Read More公众号爆款封面工坊：多尺寸封面与标题精排分离
公众号爆款封面工坊（@davtime/wechat-cover-studio，v1.0.1）是 SkillHub design-media 类 enterprise Skill，由达维发布。它面向公众号主封面 2.35:1、方形缩略 1:1 与朋友圈等多尺寸输出，内置 30+ 视觉风格家族与 17 大内容赛道库，采用视觉底图生成与标题精排分离以保障中文标题可读，并支持参考图、数据驱动方法论、A/B 变体、交互精修与自检。安装需按 skillhub.cn/install/skillhub.md 将包装入 Agent 的 skills 目录；requires_api_key 为 false。
Read More调用中转 API 生图：生图外挂.Skill 与 GPT Image 2.5 / Nano Banana 2.1
生图外挂.Skill（@user_2c58fc0f/ch12893719743826428329324，v5.2.1）是 SkillHub design-media 类社区 Skill，通过用户自备的中转 API 与 API Key，为 Agent 提供文生图、图生图与多图创作，并对接 GPT Image 2.5、Nano Banana 2.1 等模型。本文基于 SkillHub 公开页与安装文档，说明定位、已核实能力、skillhub CLI 安装命令及 Cursor 等客户端的 skills 目录注意事项，并提醒 Key 安全与中转服务依赖。未在官方页出现的配置字段不作臆测。
Read Moredsh-openrouter-imagen-v2: An Image Generation Plugin Based on OpenRouter
`dsh-openrouter-imagen-v2` is an image generation plugin for DeepSeek Harness (DSH) that leverages the OpenRouter API to provide a standalone workbench and Agent tools. It supports features such as manual or AI-generated images, reference images, model management, and seed backfill. Installation requires DSH 0.2.x, and it is ready to use after configuring the OpenRouter key. When invoked by Agent, the tool can generate a single PNG at a time and supports multiple resolutions and parameter configurations.
Read Moreimage-gen-mcp: DeepSeek Harness Image Generation Plugin
`image-gen-mcp` is a plugin for DeepSeek Harness that provides models with standardized image generation capabilities via the MCP protocol. It supports connecting to any OpenAI-compatible image generation endpoint, and once mounted, models gain access to the `image_generate` and `image_config_status` tools. Installation requires Node.js ≥ 22 and can be added via the CLI or desktop client. Configuration requires parameters such as `baseUrl` and `apiKey`, and the plugin automatically handles NDJSON-to-LSP format conversion. Built-in skills guide the model to make calls according to a workflow, making it suitable for giving models multimodal generation capabilities.
Read More`tuofangzhe/dsh-plugin-image-gen`: DeepSeek Harness Image Generation Tool
The DeepSeek Harness plugin `dsh-plugin-image-gen` provides an OpenAI-compatible image generation tool that wraps asynchronous requests into synchronous operations. It supports multi-provider configuration, writes generated images directly into the workspace, and stores references for easy access by models. The plugin must be installed via `dsh plugin` and enabled in the configuration file. Its core features include strict file naming validation, exclusive write behavior (no overwriting), and support for PNG, JPEG, and WebP formats.
Read Moredsh-generation-image: DeepSeek Harness Image Generation Plugin
`sfyyy/dsh-generation-image` is a plugin that provides image generation capabilities for DeepSeek Harness (DSH). Through the `generate_image` tool, it enables Agents to call OpenAI-compatible interfaces (such as xiaoyaoapi), persist generated images as DSH attachments, and display them in the UI. The plugin supports text generation, image editing, and multi-format data downloads. After installation, the Web process must be restarted, and users need to configure their own API keys and endpoints. It rewrites images into textual tokens, ensuring that text models do not directly handle image data. This plugin is part of the community-maintained DSH ecosystem and provides seamless multimodal extension capabilities for agent development.
Read MoreDeepSeek Harness Image Generation Tool: Integrating ToAPIs to Provide Text-to-Image and Image-to-Image Capabilities
`dsh-tool-image-gen` is an image-generation tool plugin for DeepSeek Harness, maintained by zhangjunjesse. It integrates with the ToAPIs asynchronous API and supports text-to-image, image-to-image, and transparent-background generation. The plugin offers automatic task polling, local reference-image upload, and concurrent generation of multiple images, and supports various sizes and resolutions. An API key must be configured, custom parameters are supported, and it can be invoked directly within DSH. Note that the plugin is billed through the ToAPIs platform, and some preview features are limited on Windows.
Read Moredsh-nanobananapro: NanoBananaPro Image and Video Generation Plugin for DeepSeek Harness
`dsh-nanobananapro` is a DeepSeek Harness plugin maintained by SynMind AI, designed to integrate NanoBananaPro's image and video generation capabilities directly into the DSH toolchain. The plugin provides native interfaces through a public API, enabling users to call models such as Veo 3 for content generation, while also offering balance inquiry and task management tools. It can be installed using the `npx` command and is ready for direct use in DSH conversations or workflows. Users should note that the plugin consumes only API wallet balance, does not support local asset uploads, and paid tools require manual approval by default. Timeout policies can be adjusted via configuration files.
Read MoreDeepSeek Harness Image Generation Plugin
The `dsh-image-gen` plugin, maintained by GooDAnDReaDY, is designed to deeply integrate image generation into DeepSeek Harness (DSH), replacing the need to switch to traditional external tools. Built on the FAL queue API, it supports multiple providers and allows the model to output images directly within conversations. Its core features include image variant tuning, smart aspect-ratio cropping, background removal, and style transfer. In addition, it offers project asset bundle export (e.g., PWA icons), a native sidebar gallery, and interactive diagnostic tools, providing front-end developers and designers with a one-stop solution for image processing and asset management.
Read Moredsh-codex-image-bridge: Adding Visual Capabilities to DeepSeek Harness
The DeepSeek Harness plugin `dsh-codex-image-bridge` adds visual capabilities to AI assistants. It supports image generation, detailed editing based on existing images, understanding image content, and checking service status. Installation requires the image service script to be available and the user to be logged in. Configuration options include the output directory, image generation model, timeout duration, and sandbox settings, among others. Generated or edited images are displayed in the session and saved as local files. The plugin is maintained by xmasdong and is licensed under the MIT License.
Read Moredsh-mindseye: Adding Image Understanding, Image Generation, and Visible Browser Automation to DeepSeek Harness
## Introduction In agent development, these issues frequently arise: the model is text-only, so when a user pastes a screenshot and asks, "What is this error?", the model cannot answer. To generate an image, one must manually open another tool, download the file, and drag it back into the session. To operate a webpage, mainstream solutions either run invisible headless browsers or directly take over the user's browser profile, which carries significant risks. dsh-mindseye (MindsEye) is specifically designed to address these three gaps. It is a plugin for DeepSeek Harness (DSH); in a nutshell, it provides intent-driven vision, image generation, and visible browser automation for
Read Moredsh-codex-tools: Adding Codex Search, Image Generation, and Image Recognition Tools to DeepSeek Harness
`dsh-codex-tools` is a plugin for DeepSeek Harness (DSH), maintained by SPYQWER1, providing three tools: `codex_web_search`, `image_gen`, and `image_vision`, which respectively enable public web search summarization, bitmap generation, and local image understanding. This plugin reuses existing ChatGPT/Codex login sessions and does not offer standalone login or LLM services. Search supports customizing the number of sources and freshness; image generation supports multiple sizes and formats, with output paths restricted to the DSH session workspace, prohibiting absolute paths, parent directories, and symbolic links, and not overwriting existing files; image vision recognition only accepts local files within the workspace that meet specific format and size constraints. Installation requires Node>=22 environment, supports adding via Git or npm, and after installation
Read MoreDirect Image Generation in DeepSeek Harness: Getting Started with the dsh-image-gen Plugin
dsh-image-gen is an open-source DeepSeek Harness image generation plugin by shanliuling, which supports direct image generation via the generate_image tool within web dialogs. It is compatible with multiple providers including Gemini, OpenAI, Volcano Ark Seedream, and Tongyi Wanxiang, and comes with built-in gallery management and workspace persistence. This article covers installation commands, API Key configuration, typical usage, as well as plugin permissions and security considerations.
Read MoreGenerate bitmap assets in the coding workflow using Codex's imagegen Skill
# Introduction to OpenAI Codex System-level Skill `imagegen` This skill enables generation and editing of bitmap assets including photos, illustrations, mockups and sprites within coding conversations via either the built-in `image_gen` command or an explicit CLI. Details including default auto-installation location, two operating modes, prompt templates, save path conventions, applicable boundaries and official CLI examples are provided to facilitate directly integrating visual assets into project repositories.
Read MoreExperimental Measurement of Z-Image: An Efficient Image Generation Model with 6B Parameters
Z-Image is an efficient image generation model with 6B parameters, achieving or even surpassing the performance of mainstream competitive models with 8 inference steps (8 NFEs). It can run smoothly on consumer-grade devices with 16G VRAM. The model has three variants: Turbo (lightweight and real-time, suitable for AIGC applications and mini-programs), Base (undistilled for secondary fine-tuning), and Edit (specialized for image editing), with Turbo being the most valuable for practical deployment. In practical tests, the generation time for 1024×1024 resolution is 0.8 seconds (with Flash Attention + model compilation), and the peak memory usage is 14G. Technically, its S3-DiT architecture enhances parameter efficiency, the Decoupled-DMD distillation algorithm enables 8-step inference, and DMDR fuses RL and DMD to optimize quality. Its strengths lie in bilingual text rendering, photorealistic generation, low-VRAM deployment, and image editing. Limitations include only Turbo being open, and the need for optimization in extreme stylized generation and model compilation time. Z-Image balances performance, efficiency, and practicality, making it suitable for small and medium-sized teams and developers to reduce deployment barriers.
Read More