dsh-local-model：让 DeepSeek Harness 直连本地推理服务
dsh-local-model 是 songzhiyin 维护的 DSH 模型推理插件，注册 local provider，将对话与 Agent 请求转发至本机 Ollama、LM Studio、llama.cpp、vLLM 等端点，支持 OpenAI 协议与 Ollama 原生 API，含工具调用转换与文本 JSON 兜底、模型自动发现及设置页连接测试。无需登录 DeepSeek、不消耗云端 token。安装示例：dsh plugin --profile web add dsh-local-model。要求 DSH ^0.2.0-rc.2、Node >= 20，MIT 许可。
Read Moredsh-localforge: 连接 DeepSeek Harness 与本地 LLM 服务器
`dsh-localforge` 是 DeepSeek Harness 的高性能本地模型适配器插件，支持 LM Studio 及 OpenAI/Anthropic 兼容端点。它解决了本地模型并发控制与容错难题，具备模型发现缓存、自动加载、并发限制、推理预算、采样预设及 Prometheus 监控等功能。插件通过插件机制直接调用本地算力，在 DSH Web 界面启用后，即可通过配置文件连接本地服务器，实现无需 API 余额的模型推理。
Read MoreAI批量短视频生产工作流：知识类短视频从选题到分发的 Agent 流水线
介绍 SkillHub 社区 Skill「AI批量短视频生产工作流」（@user_d922c05a/ai-video-batch-production，v1.0.0）。它按官方说明串联选题挖掘、脚本生成、画面素材、声音克隆、剪辑合成与批量分发，面向知识类博主批量出片；标签 requires_api_key=false，但官方注明仅在本地大模型环境验证。文章说明安装方式（skillhub install 与 SkillHub 安装文档）、Cursor/Claude/Codex 常见 skills 路径、触发词与典型对话用法，并提醒线上模型需自配与合规自查。
Read Moredsh-token-usage：DeepSeek Harness 本地 Token 用量统计
`dsh-token-usage` 是一个 DeepSeek Harness 客户端插件，旨在帮助开发者监控本地大模型的 Token 消耗成本。该插件不依赖第三方接口，而是直接读取本地会话日志，在设置页提供可视化的用量统计面板。核心功能包括总用量汇总、GitHub 风格的热力图日历以及按供应商分组的模型用量详情。插件采用 MIT 许可证，支持通过脚本安装与卸载，并允许通过配置文件临时停用。它确保数据安全，仅读取不修改用户数据，统计口径与 DSH 持久日志一致，适合需要精细化控制成本的开发者使用。
Read Moredsh-model-bridge: Expose DSH Models as a Local OpenAI / Anthropic Gateway
\`dsh-model-bridge\` is a plugin for DeepSeek Harness (DSH) designed to address the issues of scattered and exposed API keys when local tools call large models. It encapsulates models configured in DSH as a local OpenAI/Anthropic-compatible gateway on port 8791, enabling zero-configuration access. The plugin supports dual protocols, tool calling, and secure management of local API keys, allowing other software to use it by simply configuring the local address and API key. However, the gateway must be started while DSH is running and does not support 24/7 persistent operation.
Read Moredsh-wsl-llamacpp: A Plugin for Integrating with Local llama.cpp / Unsloth Desktop
`dsh-wsl-llamacpp` is a plugin for DeepSeek Harness (DSH) designed to simplify access to local LLM inference services. The plugin connects to local llama.cpp or Unsloth Desktop via an OpenAI-compatible interface, provides the `llama_status` and `llama_chat` tools, and supports reusing GGUF models. The installation command is `dsh plugin add github:173787247/dsh-wsl-llamacpp`. Users can configure the baseUrl, default model, and timeout. The plugin runs with the current process's permissions and is suitable for scenarios that require managing local WSL environment inference flows within DSH.
Read Moredsh-llm-subscription: Using Claude, Gemini, Grok, and Local Ollama via Subscription
`dsh-llm-subscription` is a native DeepSeek Harness plugin designed to address the challenge of accessing LLMs via subscriptions without manual API key management. It supports Claude, Gemini, and Grok subscription models, as well as local Ollama models, providing a free and convenient access experience. The plugin calls a local CLI to reshape responses into the format expected by DSH, but it is limited by the lack of incremental streaming, inability to access reasoning content, and no support for image input. Additionally, it carries compliance risks, particularly because violations involving Anthropic and Google are subject to severe penalties; Grok is the only risk-free option.
Read MoreIntroduction to the dsh-llama-model-manager Plugin
dsh-llama-model-manager is a DeepSeek Harness plugin designed specifically for the Windows environment, aimed at resolving the issues of cumbersome model switching, chaotic process management, and resource leaks during local deployment of large models. Through a fixed OpenAI-compatible gateway, it automatically manages the lifecycle of `llama-server`, including model loading, unloading, graceful shutdown, and VRAM release. It supports operations via a visual interface, provides process status monitoring and log viewing, and eliminates the need for frequent configuration changes, offering developers a stable and efficient local inference solution.
Read Moredsh-chatgpt-web: Bridging ChatGPT Web to DeepSeek Harness
The `dsh-chatgpt-web` plugin for DeepSeek Harness (DSH) provides DSH with a zero-cost, OpenAI-compatible interface by reusing free ChatGPT web sessions via local Playwright browser automation. It supports streaming, automatic model detection, and lifecycle management, but is limited to single-tab execution, has no concurrency capability, and is constrained by OpenAI free-tier limits, making it suitable only for use cases such as pure chat and code explanation.
Read Moredsh-plugin-lmstudio: DSH Plugin for Integrating Local LM Studio Models
dsh-plugin-lmstudio is a DeepSeek Harness (DSH) plugin designed to simplify the integration of local large models. Through OpenAI-compatible routing, it maps LM Studio's local inference capabilities to DSH's conversation layer, enabling local conversations without consuming cloud API quotas. Core features include using LM Studio models, the default `lm-studio` route, and local port communication. After installation, start the server in LM Studio's Developer mode and load a model, then retrieve the model ID via the terminal and configure it in the DSH configuration file, with hot reloading supported. This plugin is suitable for developers who need to run models locally in an isolated environment or avoid API rate limits.
Read Moredsh-local-llm-controller: Integrating Local llama.cpp Capabilities for DSH
The `dsh-local-llm-controller` plugin is designed to address the cumbersome process of deploying large models locally in DeepSeek Harness (DSH), enabling one-click start/stop and configuration management through its “everything is a plugin” approach. It supports A/B dual-slot management, switching between text and vision modes, and 8 parameter presets, and is automatically registered after installation. Users must prepare `llama-server` and model files themselves; after configuring the directories and paths in the plugin, they can start the service with one click. The plugin is currently compatible only with the DSH RC branch, does not include the server itself, and its vision mode requires specific build-version support.
Read Moredsh-mlx-local: Running Local Large Models on Apple Silicon Macs
`dsh-mlx-local` is a DeepSeek Harness plugin designed specifically for Apple Silicon Macs to simplify local model inference. It automatically manages Python environments, virtual environments, and the `mlx_lm.server` service, seamlessly integrating local inference into DSH’s custom provider system. Core features include model directory management, automatic environment setup, service start/stop control, and thinking-intensity optimization for Qwen3. The plugin does not register additional providers or tools and runs purely as infrastructure. Users must manually start the service and configure the Provider in settings to connect to the local API. It supports macOS 13 and later.
Read More