小米 MiMo V2-TTS 语音合成技能：在 Agent 里用 OpenAI 兼容接口做 TTS
介绍 SkillHub 社区 Skill「小米 MiMo V2-TTS 语音合成技能」（@user_512254e3/mimo-tts，v1.2.0）：通过 mimo-tts CLI 与 MIMO_API_KEY 将文本合成为 WAV/pcm16，支持预置音色与 style 风格控制。涵盖 skillhub 安装、各 Agent skills 目录、npm 依赖、典型命令示例及 API Key、文本长度与流式格式等注意事项。
Read Morenim-interactive-chat：为 NVIDIA NIM 模型生成一键交互对话脚本
nim-interactive-chat 是 SkillHub 社区 Skill（@user_93ce1c9f/nim-interactive-chat，v1.0.0），面向 AI Agent，指导根据 Docker 镜像、容器名、端口与模型名生成启动、停止及 Python 交互客户端脚本。支持 OpenAI 兼容 API、流式输出与上下文保持，自动等待健康检查，可选 GPU 显存利用率。无需 API Key。通过 skillhub install 安装到 Cursor 等客户端的 skills 目录后，使用 create-nim-interactive 即可复现文档示例流程。
Read Moredsh-omni-modal：为纯文本模型接上 OpenAI 兼容的生图与生视频
dsh-omni-modal 是 yzhclcl 维护的 DeepSeek Harness 插件，在设置侧栏提供 Omni-modal settings，可配置多套 OpenAI 兼容的图像与视频 endpoint，并向模型暴露 generate_image、generate_video、check_video。支持配置档切换、图像与视频独立字段、额外 JSON 参数与连通性测试；视频异步提交与轮询，任务可跨重启。安装命令为 dsh plugin --profile <profile> add github:yzhclcl/dsh-omni-modal，需 Node.js ≥18，安装后重启 DSH。插件以进程权限访问第三方 API 与本地 media 目录，使用前请核对源码与 MIT 许可证。
Read MoreDsh-OtherAPI：在 DSH 0.1.7 下一键接入 OpenAI 兼容 API
Dsh-OtherAPI 是 ShukebtAb 维护的 DeepSeek Harness Web 插件，基于 dsh-plugin-thinking-api 改造，适配 DSH 0.1.7-rc.2 的 entry-config 设置模型。通过设置中的 OtherAPI 面板管理 OpenAI 兼容 provider、模型启用与思考模式，密钥经 credentials 写入本地 refs 而非明文进配置。安装命令为 dsh plugin --profile web add github:ShukebtAb/Dsh-OtherAPI；插件 0.1.0 已在 0.1.7-rc.2 验证，不支持 0.1.5–0.1.6。
Read Moredsh-lantern：把本机 DSH 模型以 OpenAI / Anthropic 兼容接口开放给局域网
dsh-lantern（LANtern）是 Saretheya 维护的 DSH 插件，在本机只读复用已注册模型与上游登录态，通过 standalone 或 reuse 监听提供 /v1/chat/completions、/v1/models 等 OpenAI 兼容端点及可选 Anthropic /v1/messages。模型名须写全 <模型>@<provider>，局域网访问依赖自签 API Key、限流、熔断与本机并发保留。安装命令为 dsh plugin --profile web add dsh-lantern，需重启 DSH；支持 DSH 0.1.5 与 0.2.0，数据仅写入插件 data/ 目录。
Read Moredsh-provider-hub：在 DSH 里统一配置 BharatCode、OpenAI 兼容端点与订阅模型
dsh-provider-hub 是 arzvaak 维护的 DeepSeek Harness 模型推理类插件，在 Settings → Providers 中集成 BharatCode 预设（含模型发现与 API Key）、OpenAI 兼容端点（Chat Completions / Responses），并通过 dsh-plugin-subscriptions 嵌入 ChatGPT Codex 与 GitHub Copilot 等订阅登录。密钥写入 DSH credential store，使用原生 llm-pi-ai 适配器。安装命令为 dsh plugin --profile desktop add github:arzvaak/dsh-provider-hub，需 DSH >=0.2.0-rc.2；订阅功能需同 profile 安装 dsh-plugin-subscriptions。
Read Moredsh-model：多家模型汇成一个 OpenAI 兼容地址
dsh-model 由 AIcivilization 维护，将 Codex、Kimi、WorkBuddy、Gemini、Groq、Ollama 等多来源模型统一到本机或 VPS 上的 OpenAI 兼容 /v1 入口，一把 key 供 dsh 与任意 OpenAI 客户端使用。支持来源开关、组合 failover、多来源自动切换、每日签到与失败/用量统计。安装：npm install -g dsh-model 后执行 dsh-model setup；在 dsh 中于设置 → dsh-model 管理。需 Node ≥ 20、dsh ≥ 0.2.0-rc.2，MIT 许可。
Read Moregpt-image-2生图技能：在 Agent 里自动发现网关并落盘出图
gpt-image-2生图技能（@user_fdb96f95/gpt-image-2，v1.1.2）是 SkillHub 社区 design-media 类技能，需自备 API Key。它指导 Agent 从环境变量或 config.yaml 发现 OpenAI 兼容网关，用同一 base URL 校验 /v1/models 是否含 gpt-image-2，再调用 /images/generations，将 b64_json 解码为本地图片。文章给出 skillhub 安装命令、Cursor/Claude/Codex skills 目录及可复现的 Python 示例，并说明网关路径不一致、尺寸回退等注意点。
Read Moredsh-plugin-thinking-api
`dsh-plugin-thinking-api` is an open-source plugin for DeepSeek Harness, designed to resolve the issue where third-party OpenAI-compatible endpoints (such as Tencent CodeBuddy) do not support the official `developer` role, causing reasoning requests to fail. The plugin avoids compatibility conflicts by forcing the use of the `system` role, supports one-click configuration of the generic OpenAI format, and automatically enables thinking mode. Users can install it by editing `package.json` and configure it using YAML settings or the Web UI wizard. It is suitable for scenarios that require enabling reasoning capabilities on non-official endpoints.
Read Moredsh-plugin-vision: DeepSeek Harness Vision Assistant Plugin
The `dsh-plugin-vision` plugin for DeepSeek Harness (DSH) addresses the lack of vision capabilities in the official API by registering the `vision_describe` tool. It enables the main model to call any OpenAI-compatible vision API to describe images. Installation must be performed using the official script, and the plugin supports configuration of custom API endpoints, API keys, and caching strategies. It relies on the global dependency tree, so the `dsh plugin add` command must not be used during installation. Additionally, when used with DSH Desktop, it can automatically convert pasted images to file paths, provided that dependencies are installed correctly.
Read Moredsh-openai-gateway: Expose DSH Agent as an OpenAI Compatible API
`dsh-openai-gateway` is a plugin for DSH designed to expose the DeepSeek Harness (DSH) Agent as a standard OpenAI-compatible API server. The plugin supports endpoints such as `POST /v1/chat/completions` and `GET /v1/models`; each request is backed by a real DSH Agent session equipped with tools and a workspace. Users simply need to install the plugin and restart the service to connect using any OpenAI client by configuring the `base_url` and API key. By default, the plugin listens on the local loopback address port 41540 and offers two session modes: stateless and sticky. Sticky sessions support multi-turn continuous tasks and require specific headers or fields to maintain the session state.
Read Moredsh-plugin-codex-bridge: Bridge OpenAI Codex skills, instructions, and configuration to DeepSeek H
dsh-plugin-codex-bridge is a DSH plugin maintained by YYTbit and based on the MIT license, designed to address the difficulty of reusing OpenAI Codex local files. The plugin implements "zero migration" by automatically bridging Codex's skills, instructions, and configuration to DeepSeek Harness, thereby avoiding the need for users to manually move or adjust scattered files. Specific features include injecting files under ~/.codex/skills as available skills, injecting instructions.md as system prompts, and injecting model context from config.toml. Users can install it in a specified profile using the `dsh` command.
Read Moredsh-draw-router: Add unified drawing routing to DeepSeek Harness
`dsh-draw-router` is a unified drawing routing plugin for DeepSeek Harness (DSH), designed to address the inconsistencies of interfaces across different OpenAI-compatible image generation services. By registering the two Agent tools `draw_image` and `draw_list_sources`, the plugin allows DSH Agents to directly invoke drawing capabilities during conversations. It also supports the automatic discovery of available image models without relying on a fixed model list. Compatible with multiple providers such as SenseNova, StepFun, Qwen, and Hunyuan, it only requires specifying the API Root during configuration to automatically adapt paths. In addition to Agent integration, the plugin provides a REST API for managing drawing sources and initiating drawing requests.
Read Moredeepseek-vl-support: DSH Plugin for Integrating Vision into DeepSeek Text Models
`deepseek-vl-support` is an MIT-licensed plugin designed to address the pain point where DeepSeek text models cannot directly read images. The plugin intercepts image reading requests, forwards them to selected OpenAI-compatible vision services (such as OpenRouter, SiliconFlow, DashScope, etc.), and returns the text descriptions in a fixed format to the model, thereby enabling image understanding. This tool supports multiple Agent environments, including Claude Code, Codex, DeepSeek Harness, Cursor, VS Code, etc., and provides a one-time installation wizard, per-Agent installation commands, and a `doctor` health check function. Users only need to have Node.js...
Read Moredsh-polyglot: Providing OpenAI-Compatible Endpoint Switching and Automatic Fallback for DSH
dsh-polyglot is an MIT-licensed DSH plugin developed by Jesse-njx, designed to address challenges in DSH when invoking external models, including multi-endpoint management, unstable free quotas, and rate limiting. This plugin consolidates multiple OpenAI-compatible providers into a single virtual provider, `polyglot`, utilizing a universal adapter to handle streaming, tool calls, and usage extraction, while employing declarative configuration to accommodate variations across different services. At its core is an automatic fallback router that, upon encountering 429 rate limits, quota exhaustion, 5xx errors, or missing API keys, automatically switches to the next available provider in the chain, implementing exponential backoff and retry strategies for cooldown. The plugin maintains provider information through JSON preset files and records the service source, status, and token usage for each request in session logs.
Read Moredsh-llm-newapi: Adding NewAPI Provider to DeepSeek Harness
dsh-llm-newapi is an LLM Provider plugin based on the MIT license, designed to add a NewAPI provider to DeepSeek Harness (dsh), enabling integration with OpenAI-compatible gateways without any code modifications. The plugin implements the `LlmAdapter` interface, utilizes the `POST /chat/completions` and `GET /models` endpoints to provide model services, and automatically filters out models that only support chat functionality. Users can install the plugin via npm or GitHub, and need to add the package name to the `bundles` array in the dsh web configuration, then restart the service. The API Key is not read from environment variables and must be configured through the settings page of dsh web.
Read MoreAdd DeepSeek V4 protocol adaptation for OpenAI client using deepseek-harness
HenryZ838978-maintained deepseek-harness provides protocol-aware adaptation for DeepSeek V4-Pro / V4-Flash, recording 16 protocol features, 10 contracts and 12 probes. This article verifies the access methods including pip, dsh CLI, MCP and SKILL.md, as well as precautions such as reasoning_content 400, prefix caching and the same-name conflict with the official dsh, based on the community directory and GitHub README.
Read More