{
  "site": "纳雪",
  "updated_at": "2026-10-15",
  "scope": "Configuration recipes, not a compatibility certification. No real credentials.",
  "clients": [
    {
      "id": "hermes-agent",
      "name": "Hermes Agent",
      "base_url": "https://api.xxxiai.com/v1",
      "steps": [
        "退出当前聊天，在终端运行 hermes model，选择 Custom endpoint。",
        "Base URL 填 https://api.xxxiai.com/v1，填写纳雪 Key 与完整模型 ID。",
        "若向导提供 API mode，聊天分组选 Chat Completions；GPT 编程线路按其支持情况选 Responses。保存后启动新会话测试。"
      ],
      "notes": "聊天中的 /model 用于切换已配置模型，不能代替新增服务商向导。压缩、视觉等辅助任务也要使用有权限的模型。",
      "source": "https://hermes-agent.nousresearch.com/docs/integrations/providers",
      "verification": "configuration guidance; end-to-end client not tested in this change"
    },
    {
      "id": "cursor",
      "name": "Cursor",
      "base_url": "https://api.xxxiai.com/v1",
      "steps": [
        "打开 Cursor Settings → Models，找到 API Keys。",
        "只有当前版本提供 Override OpenAI Base URL 等自定义入口时，才把地址改为 https://api.xxxiai.com/v1，并填写纳雪 Key。",
        "添加或启用有权限的模型，在新对话中先测试普通问答。"
      ],
      "notes": "自带 Key 与任意第三方地址支持不是一回事；没有自定义地址入口的版本不能按此接入。Tab 补全等功能不一定使用自定义 Key，且 Cursor 的请求会经过其后端。",
      "source": "https://cursor.com/help/models-and-usage/api-keys",
      "verification": "configuration guidance; end-to-end client not tested in this change"
    },
    {
      "id": "workbuddy",
      "name": "WorkBuddy",
      "base_url": "https://api.xxxiai.com/v1",
      "steps": [
        "在设置中进入模型配置，添加模型并选择 自定义 / Custom。",
        "按 URL 字段用途填写：基础地址用 https://api.xxxiai.com/v1；若要求完整聊天端点，填 https://api.xxxiai.com/v1/chat/completions。",
        "填写 Key 和模型名称，保存后选为当前模型，发送简短消息。"
      ],
      "notes": "选择支持 Chat Completions 和所需工具功能的分组；界面的能力选项须按实际模型填写。",
      "source": "https://www.codebuddy.cn/docs/workbuddy/From-Beginner-to-Expert-Guide/Function-Description/Model",
      "verification": "configuration guidance; end-to-end client not tested in this change"
    },
    {
      "id": "deepseek-harness",
      "name": "DeepSeek Harness",
      "base_url": "https://api.xxxiai.com/v1",
      "steps": [
        "在已运行的 Web UI 中进入设置 → 模型 → 添加模型提供商，选择自定义模型 API。",
        "使用兼容的 OpenAI 协议，Base URL 填 https://api.xxxiai.com/v1，Key 填纳雪密钥。",
        "添加完整模型 ID，保存、启用并新建测试会话。若版本支持协议选择，确保它与所选分组一致。"
      ],
      "notes": "供应商名称不限定只能用 DeepSeek；可用范围由纳雪 Key 的分组决定。先验证文字，再验证编程工具。",
      "source": "https://github.com/deepseek-ai/deepseek-harness/blob/main/docs/user/guide/providers.zh.md",
      "verification": "configuration guidance; end-to-end client not tested in this change"
    },
    {
      "id": "zcode",
      "name": "ZCode",
      "base_url": "https://api.xxxiai.com/v1",
      "steps": [
        "打开模型选择器，进入管理模型 / 模型设置，然后添加供应商。",
        "选择当前版本提供的自定义或 OpenAI 兼容入口，填入 https://api.xxxiai.com/v1、纳雪 Key 和完整模型 ID。",
        "保存后在新会话选择该供应商，先请求简短回复。"
      ],
      "notes": "不同版本可选的 provider 和字段可能改变；如无自定义地址入口，先核对官方说明。",
      "source": "https://zcode.z.ai/cn/docs/configuration",
      "verification": "configuration guidance; end-to-end client not tested in this change"
    },
    {
      "id": "cline",
      "name": "Cline",
      "base_url": "https://api.xxxiai.com/v1",
      "steps": [
        "进入 VS Code 的 Cline 面板 → 设置 → API Provider，选择 OpenAI Compatible。",
        "Base URL 填 https://api.xxxiai.com/v1，API Key 填自己的 Key，Model ID 填完整模型 ID。",
        "先测试普通问答，成功后在独立测试目录验证读取文件等工具。"
      ],
      "notes": "上下文、图片能力、价格和输出限制按真实模型配置；支持文字不代表支持全部 Agent 功能。",
      "source": "https://docs.cline.bot/provider-config/openai-compatible",
      "verification": "configuration guidance; end-to-end client not tested in this change"
    },
    {
      "id": "continue",
      "name": "Continue",
      "base_url": "https://api.xxxiai.com/v1",
      "steps": [
        "打开 Continue 的本地配置文件 config.yaml，合并下方模板。",
        "通过 Continue 的 secrets 功能配置 NAXUE_API_KEY，将 MODEL_ID 换成真实模型。",
        "重新加载配置，在聊天中选中“纳雪”，发送一条短消息。"
      ],
      "notes": "模板明确使用 Chat Completions；仅对已确认支持 Responses 的分组与客户端配置启用 useResponsesApi。不要把聊天模型用作 embedding 或 rerank。",
      "source": "https://docs.continue.dev/customize/model-providers/top-level/openai",
      "verification": "configuration guidance; end-to-end client not tested in this change"
    },
    {
      "id": "gemini-cli",
      "name": "Gemini CLI（待原生接口验收）",
      "base_url": null,
      "steps": [
        "先确认用途：Gemini CLI 需要 Google 原生 generateContent 协议，不能直接使用 OpenAI 的 /v1/chat/completions。",
        "当前纳雪公网 /v1beta 路径尚未接通验收；控制台能生成导入配置，不代表原生请求已可用。",
        "现阶段使用 Cherry Studio、Chatbox 等兼容客户端，通过 https://api.xxxiai.com/v1 调用对应 Gemini 分组。"
      ],
      "notes": "本页暂不提供看似可直接运行的 Gemini CLI 配置；原生线路完成验收后再更新。",
      "source": "https://geminicli.com/docs/get-started/authentication/",
      "verification": "configuration guidance; end-to-end client not tested in this change",
      "status": "pending native endpoint validation"
    },
    {
      "id": "opencode",
      "name": "OpenCode",
      "base_url": "https://api.xxxiai.com/v1",
      "steps": [
        "在 OpenCode 配置中添加自定义 provider；下方模板使用 @ai-sdk/openai-compatible。",
        "在启动它的环境中设置 NAXUE_API_KEY，替换模型占位符。",
        "打开 OpenCode 并选择 naxue/MODEL_ID，先测试文字，再测试工具。"
      ],
      "notes": "本模板面向 Chat Completions 分组。Responses 线路需使用明确支持 Responses 的 provider 配置，不能只替换模型名。",
      "source": "https://opencode.ai/docs/providers/",
      "verification": "configuration guidance; end-to-end client not tested in this change"
    },
    {
      "id": "aider",
      "name": "Aider",
      "base_url": "https://api.xxxiai.com/v1",
      "steps": [
        "安装 Aider 后，在启动终端设置 OPENAI_API_BASE=https://api.xxxiai.com/v1。",
        "将 OPENAI_API_KEY 设置成纳雪 Key，运行 aider --model openai/MODEL_ID。",
        "把 MODEL_ID 换成完整 ID，在空测试目录先发简短问题，再尝试小范围文件修改。"
      ],
      "notes": "使用 Chat Completions 兼容分组。未知模型可能需要额外的上下文长度、费用或编辑格式配置。",
      "source": "https://aider.chat/docs/llms/openai-compat.html",
      "verification": "configuration guidance; end-to-end client not tested in this change"
    },
    {
      "id": "dify",
      "name": "Dify",
      "base_url": "https://api.xxxiai.com/v1",
      "steps": [
        "在模型供应商中安装或选择 OpenAI-API-compatible 插件。",
        "新增 LLM 模型，填写完整模型名、Endpoint URL https://api.xxxiai.com/v1 和纳雪 Key。",
        "保存并验证，在测试应用中选择该模型后发消息。"
      ],
      "notes": "实际请求由 Dify 服务端发起，服务端需要能访问纳雪。知识库的向量模型应另外配置。",
      "source": "https://github.com/langgenius/dify-official-plugins/tree/main/models/openai_api_compatible",
      "verification": "configuration guidance; end-to-end client not tested in this change"
    },
    {
      "id": "n8n",
      "name": "n8n",
      "base_url": "https://api.xxxiai.com/v1/chat/completions",
      "steps": [
        "添加 HTTP Request 节点，方法选 POST，URL 用 https://api.xxxiai.com/v1/chat/completions。",
        "使用 Header Auth 凭据：名称 Authorization，值为 Bearer  后接你的 Key。",
        "请求体选择 JSON，填 model、messages 和 stream:false；先手动执行一个测试节点。"
      ],
      "notes": "Key 保存在 n8n Credentials 中。请求超时后先查用量，避免自动重试导致重复生成。",
      "source": "https://docs.n8n.io/integrations/builtin/core-nodes/n8n-nodes-base.httprequest/",
      "verification": "configuration guidance; end-to-end client not tested in this change"
    },
    {
      "id": "http-tools",
      "name": "curl / Postman / Apifox",
      "base_url": "https://api.xxxiai.com/v1",
      "steps": [
        "新建请求，或导入本页 API 示例 / OpenAPI 文件。",
        "配置私有环境变量 NAXUE_API_KEY，认证选择 Bearer Token；Messages 示例也支持 x-api-key。",
        "选择正确的 POST 接口，Body 用 JSON，先用最小请求测试。"
      ],
      "notes": "使用完整接口地址，不能只请求根域名。分享集合或导出文件前移除真实凭据。",
      "source": "https://learning.postman.com/docs/getting-started/first-steps/sending-the-first-request/",
      "verification": "configuration guidance; end-to-end client not tested in this change"
    },
    {
      "id": "open-webui",
      "name": "Open WebUI",
      "base_url": "https://api.xxxiai.com/v1",
      "steps": [
        "管理员进入 Connections / 连接设置，新增 OpenAI 兼容连接。",
        "URL 填 https://api.xxxiai.com/v1，填写纳雪 Key，保存并获取模型。",
        "在新聊天中选中模型，发送短消息；若连接受限，检查服务端出站网络。"
      ],
      "notes": "需要管理员权限，菜单位置随版本变化；文字连接不会自动启用图片接口。",
      "source": "https://docs.openwebui.com/getting-started/quick-start/connect-a-provider/",
      "verification": "configuration guidance; end-to-end client not tested in this change"
    },
    {
      "id": "anythingllm",
      "name": "AnythingLLM",
      "base_url": "https://api.xxxiai.com/v1",
      "steps": [
        "进入 LLM Preference / 模型设置，寻找 Generic OpenAI。",
        "填写 https://api.xxxiai.com/v1、纳雪 Key 和完整模型 ID。",
        "保存并在测试工作区验证普通对话，再单独配置知识库所用的向量服务。"
      ],
      "notes": "模型上下文限制按实际分组填写；聊天模型不能代替 embedding 模型。以安装版本的字段为准。",
      "source": "https://docs.anythingllm.com/",
      "verification": "configuration guidance; end-to-end client not tested in this change"
    },
    {
      "id": "comfyui",
      "name": "ComfyUI",
      "base_url": "https://api.xxxiai.com/v1/images/generations",
      "steps": [
        "选用明确支持自定义 OpenAI Images 接口的可信节点；仅安装 ComfyUI 本体不会自动连接纳雪。",
        "若节点要求完整 URL，填 https://api.xxxiai.com/v1/images/generations，使用图片分组 Key。",
        "按节点说明填写图片模型、提示词及兼容尺寸，确认费用后再运行一张图。"
      ],
      "notes": "节点协议和响应格式必须匹配 URL / Base64 的实际返回。本页没有指定或验证第三方节点，不能承诺任意工作流可用。",
      "source": "https://docs.comfy.org/",
      "verification": "configuration guidance; end-to-end client not tested in this change",
      "status": "requires a compatible node"
    },
    {
      "id": "codex",
      "name": "Codex",
      "base_url": "https://api.xxxiai.com/v1",
      "protocol": "Responses",
      "guide": "https://api.xxxiai.com/status/guide.html#codex",
      "notes": "Use your own Naxue key and a model permitted by its group."
    },
    {
      "id": "claude-code",
      "name": "Claude Code",
      "base_url": "https://api.xxxiai.com",
      "protocol": "Messages",
      "guide": "https://api.xxxiai.com/status/guide.html#claude",
      "notes": "Use your own Naxue key and a model permitted by its group."
    },
    {
      "id": "cherry",
      "name": "Cherry Studio",
      "base_url": "https://api.xxxiai.com/v1",
      "protocol": "OpenAI compatible",
      "guide": "https://api.xxxiai.com/status/guide.html#clients",
      "notes": "Use your own Naxue key and a model permitted by its group."
    },
    {
      "id": "chatbox",
      "name": "Chatbox",
      "base_url": "https://api.xxxiai.com/v1",
      "protocol": "OpenAI compatible",
      "guide": "https://api.xxxiai.com/status/guide.html#clients",
      "notes": "Use your own Naxue key and a model permitted by its group."
    }
  ]
}