@streamapp/stream 0.0.1

package/resources/packages/reddit/reddit-sub.recipe.json

@@ -0,0 +1,77 @@
+{
+  "version": 2,
+  "kind": "fetch",
+  "sourceId": "reddit-sub",
+  "cookieDomain": "reddit.com",
+  "entryUrl": "https://www.reddit.com/r/{subreddit}/",
+  "entryWait": "domcontentloaded",
+  "request": {
+    "url": "https://www.reddit.com/r/{subreddit}/hot.json?limit=25&after={cursor}",
+    "method": "GET",
+    "headers": {
+      "accept": "application/json"
+    }
+  },
+  "pagination": {
+    "mode": "cursor",
+    "itemsAt": "data.children",
+    "cursorFrom": "data.after",
+    "cursorParam": "after",
+    "hasMore": "data.after",
+    "maxPages": 3
+  },
+  "assert": [
+    {
+      "path": "data.children",
+      "desc": "hot.json 返回 Listing data.children 数组 —— 登录墙 / IP 封锁 / 端点已变"
+    }
+  ],
+  "mapping": {
+    "guid": "data.id",
+    "title": "data.title",
+    "link": "https://www.reddit.com{data.permalink}",
+    "author": "data.author",
+    "like_count": "data.score",
+    "comment_count": "data.num_comments",
+    "pubDate": "data.created_utc",
+    "description": "data.selftext"
+  },
+  "meta": {
+    "type": "post",
+    "description": "Reddit — subreddit 热门帖流",
+    "topics": [
+      "reddit",
+      "subreddit",
+      "discussion",
+      "社区"
+    ],
+    "categories": [
+      "social-media"
+    ],
+    "facility": {
+      "key": "reddit",
+      "label": "Reddit"
+    },
+    "capabilities": [
+      "timeline"
+    ],
+    "key_param": "subreddit",
+    "priority": 80,
+    "discoverable": true,
+    "cadence_hint_seconds": 1800,
+    "auth": {
+      "type": "none"
+    },
+    "homepage": "reddit.com",
+    "params_schema": {
+      "subreddit": {
+        "type": "string",
+        "required": true,
+        "description": "subreddit 名称（如 selfhosted、LocalLLaMA）"
+      }
+    },
+    "radar": [
+      "reddit.com/r/:subreddit"
+    ]
+  }
+}

package/resources/packages/replay/README.md

@@ -0,0 +1,7 @@
+# Recipe
+
+登录态浏览器的 recipe 兜底采集引擎。
+
+无独立 facility：签名 / 登录门禁站点（小红书、雪球…）的 recipe 源都归在这个插件下。
+源来自 `recipes/` 包，这里只补插件级 name/description（manifest 是元数据唯一的家）。
+无 backend → 进程内（用户自己的 Chrome + ext-cdp 中继），不生成容器。

package/resources/packages/replay/package.json

@@ -0,0 +1,13 @@
+{
+  "name": "@streamapp/replay",
+  "version": "1.0.0",
+  "private": true,
+  "type": "module",
+  "stream": {
+    "id": "replay",
+    "name": "Recipe",
+    "tagline": "按 recipe 在你登录态的浏览器里采集",
+    "description": "把签名或登录门禁、直连与 RSSHub 都取不到的站点写成一份 recipe，在你自己登录态的浏览器里照着跑——页面自签请求、DOM 抓取，或调用站点自身的请求客户端。采集阶梯的最后一档兜底，运行期零 token。",
+    "normalizer": "generic"
+  }
+}

package/resources/packages/rsshub/README.md

@@ -0,0 +1,16 @@
+# RSSHub
+
+Upstream tagline: "Everything is RSSible".
+
+`required: true` —— 核心插件：长尾 RSS 目录的唯一来源，产品离了它就没内容。UI 上开关锁定为开、
+不可关闭。
+
+## Runtime form
+
+RSSHub runs in a WORKER THREAD, not a container and not in-process. It's the one plugin whose fetch
+logic is foreign heavyweight code (3000+ routes, a global-fetch monkeypatch, on-first-hit route
+compilation) — a worker gives it its own globals + event loop so none of that leaks into Stream,
+while keeping cookie hot-reload a postMessage away. Wiring: `src/rsshub-adapter.ts` →
+`src/rsshub-client.ts` (spawns) → `src/rsshub-worker.ts` (harness runs RSSHub). No `backend:` block:
+there is no container. This runtime isn't a declared descriptor field yet (RSSHub is its sole user;
+generalising a `worker:` schema for one occupant would be premature) — see rsshub-static-catalog.

package/resources/packages/rsshub/manifests.yaml

@@ -0,0 +1,107 @@
+# bilibili-dynamic / bilibili-following retired 2026-07-04 — the RSSHub catalog twins
+# (rsshub:bilibili/user/dynamic/*, rsshub:bilibili/followings/dynamic/*) now derive
+# cookie auth from features.requireConfig via NS_COOKIE_DOMAIN in src/rsshub-catalog.ts.
+- id: btbtla-detail-lazy
+  adapter: rsshub
+  description: BTBTLA | 指定剧名 — 影视剧种子搜索（lazy 模式：不在搜索时解析磁力，点开单条时由 resolve.ts 直连 /tdown/ 现解，所以搜索很快）。
+  topics: [影视, 电影, 电视剧, 剧集, btbtla, bt, 种子, magnet, 下载, video, tv, movie]
+  categories: [video]
+  example_queries:
+    - btbtla 搜 雍正王朝
+    - 找一下 庆余年 的种子
+    - 上载新生 下载
+  capabilities: [search]
+  auth:
+    type: none
+  route: /btbtla/detail/:name/lazy
+  cadence_hint_seconds: 1800
+  params_schema:
+    name:
+      type: string
+      required: true
+# lizhi-user / xiaoyuzhou-podcast retired 2026-07-04 — the podcast-feed Provider row
+# now expands to the RSSHub catalog twins (rsshub:lizhi/user/:id,
+# rsshub:xiaoyuzhou/podcast/:id) via {mode:'auto', matches} on their radar patterns.
+- id: rsshub-raw
+  adapter: rsshub
+  description: Raw RSSHub route passthrough for ad-hoc fetches (escape hatch).
+  topics: []
+  example_queries: []
+  capabilities: [timeline]
+  auth:
+    type: none
+  discoverable: false
+  cadence_hint_seconds: 3600
+  params_schema:
+    route:
+      type: string
+      required: true
+
+# —— 影视频道（video variant）榜单源 ——
+# 固定 route（无占位符）+ normalizer: movie —— movie normalizer 按 id 分派解析
+# rating/year/genre（见 src/content/movie.ts）。是 default-video 频道 7 条 stream 的成员。
+- id: movie-douban-playing
+  adapter: rsshub
+  normalizer: movie
+  mode: collection
+  description: 豆瓣电影 — 正在热映
+  topics: [影视, 电影, 豆瓣, 热映, movie]
+  categories: [video]
+  capabilities: [timeline]
+  facility: { key: douban, label: 豆瓣 }
+  auth:
+    type: none
+  route: /douban/movie/playing
+  cadence_hint_seconds: 21600
+- id: movie-douban-weekly
+  adapter: rsshub
+  normalizer: movie
+  mode: collection
+  description: 豆瓣电影 — 一周口碑榜
+  topics: [影视, 电影, 豆瓣, 口碑, 榜单, movie]
+  categories: [video]
+  capabilities: [timeline]
+  facility: { key: douban, label: 豆瓣 }
+  auth:
+    type: none
+  route: /douban/movie/weekly
+  cadence_hint_seconds: 21600
+- id: movie-tmdb-trend-movie
+  adapter: rsshub
+  normalizer: movie
+  mode: collection
+  description: TMDB — 本周热门电影趋势
+  topics: [影视, 电影, tmdb, trending, movie]
+  categories: [video]
+  capabilities: [timeline]
+  facility: { key: themoviedb, label: TMDB }
+  auth:
+    type: none
+  route: /themoviedb/trending/movie/week/zh-CN
+  cadence_hint_seconds: 21600
+- id: movie-tmdb-trend-tv
+  adapter: rsshub
+  normalizer: movie
+  mode: collection
+  description: TMDB — 本周热门剧集趋势
+  topics: [影视, 剧集, tmdb, trending, tv]
+  categories: [video]
+  capabilities: [timeline]
+  facility: { key: themoviedb, label: TMDB }
+  auth:
+    type: none
+  route: /themoviedb/trending/tv/week/zh-CN
+  cadence_hint_seconds: 21600
+- id: movie-imdb-popular
+  adapter: rsshub
+  normalizer: movie
+  mode: collection
+  description: IMDb — 最热门电影
+  topics: [影视, 电影, imdb, popular, movie]
+  categories: [video]
+  capabilities: [timeline]
+  facility: { key: imdb, label: IMDb }
+  auth:
+    type: none
+  route: /imdb/chart/moviemeter
+  cadence_hint_seconds: 21600

package/resources/packages/rsshub/package.json

@@ -0,0 +1,21 @@
+{
+  "name": "@streamapp/rsshub",
+  "version": "1.0.0",
+  "private": true,
+  "type": "module",
+  "stream": {
+    "id": "rsshub",
+    "required": true,
+    "name": "RSSHub",
+    "tagline": "世界上最大的 RSS 网络",
+    "description": "网易云、Twitter、Instagram、Telegram、哔哩哔哩、知乎、微博、GitHub、YouTube 等网站和应用的信息源。",
+    "homepage": "https://rsshub.app",
+    "repository": "https://github.com/DIYgod/RSSHub",
+    "docsUrl": "https://docs.rsshub.app",
+    "normalizer": "rsshub",
+    "sourceGrouping": {
+      "enabled": true,
+      "resolver": "adapter.groupByNamespace"
+    }
+  }
+}

package/resources/packages/telegram/package.json

@@ -0,0 +1,15 @@
+{
+  "name": "@streamapp/telegram",
+  "version": "1.0.0",
+  "author": "stream",
+  "keywords": [
+    "stream-recipe"
+  ],
+  "stream": {
+    "type": "recipe",
+    "name": "Telegram",
+    "tagline": "在指定频道内搜关键词",
+    "facility": "telegram",
+    "schemaVersion": 1
+  }
+}

package/resources/packages/telegram/telegram-search.recipe.json

@@ -0,0 +1,125 @@
+{
+  "version": 7,
+  "kind": "desktop",
+  "sourceId": "telegram-search",
+  "app": {
+    "process": "Telegram.exe",
+    "windowClass": "MainWindow"
+  },
+  "steps": [
+    {
+      "kind": "invoke",
+      "query": {
+        "role": "Button",
+        "name": "取消搜索"
+      },
+      "optional": true
+    },
+    {
+      "kind": "wait",
+      "ms": 400
+    },
+    {
+      "kind": "invoke",
+      "query": {
+        "role": "ListItem",
+        "nameContains": "{channel}"
+      }
+    },
+    {
+      "kind": "wait",
+      "ms": 800
+    },
+    {
+      "kind": "invoke",
+      "query": {
+        "role": "Button",
+        "name": "搜索消息"
+      }
+    },
+    {
+      "kind": "type",
+      "text": "{q}",
+      "query": {
+        "role": "Edit",
+        "name": "搜索",
+        "className": "class Ui::InputField::Inner"
+      }
+    },
+    {
+      "kind": "wait",
+      "ms": 2000
+    },
+    {
+      "kind": "invoke",
+      "query": {
+        "role": "ListItem",
+        "nameContains": "频道, {channel}"
+      }
+    },
+    {
+      "kind": "wait",
+      "ms": 2500
+    }
+  ],
+  "observer": {
+    "itemQuery": {
+      "role": "ListItem",
+      "nameContains": "{q}"
+    },
+    "fields": {
+      "content": {
+        "read": "name"
+      }
+    },
+    "dedupeBy": "content"
+  },
+  "map": {
+    "content": {
+      "from": "content",
+      "match": "名称：[\\s\\S]+"
+    },
+    "title": {
+      "from": "content",
+      "match": "名称：(.+)"
+    },
+    "description": {
+      "from": "content",
+      "match": "描述：([\\s\\S]*?)(?:\\n\\n|$)"
+    },
+    "link": {
+      "from": "content",
+      "match": "https?://[^\\s]+"
+    }
+  },
+  "read": {
+    "dedupeBy": "link",
+    "targetCount": 30
+  },
+  "meta": {
+    "description": "在指定的 Telegram 频道里搜关键词：先复位（上一轮可能停在搜索模式）→ 从左栏会话列表按名字点进频道 → 频道内搜索 → 点第一条命中让主列表跳过去 → 从主列表读完整正文（搜索结果本身只有摘要，网盘链接被截断了）。桌面 recipe（a11y 词汇，Windows UIA 后端）。打字走 ValuePattern 写进输入框（不经键盘、不需要前台、锁屏也生效，已实测），所以 `type` 带 `query`——**必须指到里层的 `Ui::InputField::Inner`**，写外层容器 `Ui::InputField` 是收不下的。仍有一处会浮到前台：invoke「搜索消息」时 Telegram 自己 activate 窗口（不是 Stream 抢的）。实测见 2026-08-04-desktop-background-typing-design.md。",
+    "normalizer": "telegram",
+    "_why_member_timeout": "桌面采集不是网络请求:点开窗口、等界面渲染、逐步点进去,一趟约 12s。按流式路统一的 15s 量会贴着边被砍(而被砍的表现只是「这个源永远搜不到」,不报错)。按源申报 25s,只慢自己这一格。上限必须小于流式路的整体预算 DEFAULT_BUDGET_MS,否则预算先到,配多少都不生效。",
+    "member_timeout_ms": 25000,
+    "capabilities": [
+      "search"
+    ],
+    "key_param": "q",
+    "cadence_hint_seconds": 3600,
+    "discoverable": false,
+    "_why_pick_in": "只在 Provider 成员面挑得到，不在频道订阅面——按频道搜的一条腿",
+    "pick_in": ["provider"],
+    "params_schema": {
+      "q": {
+        "type": "string",
+        "required": true,
+        "description": "搜索词"
+      },
+      "channel": {
+        "type": "string",
+        "required": true,
+        "description": "要搜哪个频道（左栏会话列表里的名字，包含匹配即可）。必填：不指定就等于「搜你此刻恰好开着的那个会话」，而定时轮和聚合调用都指望不上这一点。"
+      }
+    }
+  }
+}

package/resources/packages/tencent/README.md

@@ -0,0 +1,11 @@
+
+## 说明（迁自 manifests.yaml 注释）
+
+腾讯视频（v.qq.com）— 纯 http recipe（pbaccess.video.qq.com）。
+episode：GetPageData 走 page_id:vsite_episode_list，明文 JSON、无签名，UA/Origin 均非必需（不同于
+该站 search 端点——那个端点缺这两个头会被拒 20607，此处不需要）。page_size=100 通常一页拿全；
+compute.decode 把 module_list_datas 摊平成 items[]，并把 module_params.has_next（字符串
+"true"/"false"）转成真 boolean 喂 pagination.hasMore——不转的话原始字符串 "false" 在 JS 里是
+truthy，increment 引擎 `!getPath(...)` 那条 hasMore 判断会被绕过去，永远不停。page_num 超出真实
+页数时服务器会绕回第 0 页而不是给空数组，所以必须靠这个 boolean 兜底停，不能只指望"空数组=
+自然结束"那条内置规则。

package/resources/packages/tencent/package.json

@@ -0,0 +1,15 @@
+{
+  "name": "@streamapp/tencent",
+  "version": "1.0.0",
+  "author": "stream",
+  "keywords": [
+    "stream-recipe"
+  ],
+  "stream": {
+    "type": "recipe",
+    "name": "腾讯视频",
+    "tagline": "剧集 / 综艺的分集列表",
+    "facility": "tencent",
+    "schemaVersion": 1
+  }
+}

package/resources/packages/tencent/tencent-cn-episode.recipe.json

@@ -0,0 +1,98 @@
+{
+  "version": 1,
+  "kind": "http",
+  "sourceId": "tencent-cn-episode",
+  "request": {
+    "url": "https://pbaccess.video.qq.com/trpc.universal_backend_service.page_server_rpc.PageServer/GetPageData?video_appid=3000010&vplatform=2&vversion_name=8.2.95",
+    "method": "POST",
+    "headers": {
+      "content-type": "application/json",
+      "referer": "https://v.qq.com/x/cover/{cid}.html"
+    },
+    "body": "{\"page_params\":{\"req_from\":\"web_vsite\",\"page_id\":\"vsite_episode_list\",\"page_type\":\"detail_operation\",\"id_type\":\"1\",\"page_size\":\"100\",\"cid\":\"{cid}\",\"vid\":\"\",\"lid\":\"\",\"page_num\":\"{page_num}\",\"page_context\":\"\",\"detail_page_type\":\"1\",\"req_type\":\"6\",\"req_from_second_type\":\"\",\"year\":\"\",\"tab_type\":\"\",\"chapter_name\":\"\",\"is_nocopyright\":\"false\"},\"has_cache\":1}"
+  },
+  "compute": {
+    "capabilities": [],
+    "prefetch": [
+      {
+        "as": "coverPage",
+        "parse": "text",
+        "request": {
+          "url": "https://m.v.qq.com/x/m/play?cid={cid}",
+          "method": "GET",
+          "headers": {
+            "user-agent": "Mozilla/5.0 (iPhone; CPU iPhone OS 16_0 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/16.0 MicroMessenger/8.0.0"
+          }
+        }
+      }
+    ],
+    "decode": "(() => { const mld = (input.body && input.body.data && input.body.data.module_list_datas) || []; const mod = mld[0] && mld[0].module_datas && mld[0].module_datas[0]; const items = (mod && mod.item_data_lists && mod.item_data_lists.item_datas) || []; const hasNext = !!(mod && mod.module_params && mod.module_params.has_next === 'true'); const m = /itemprop=\"image\" content=\"([^\"]*)\"/.exec((input.pre && input.pre.coverPage) || ''); const cover = m ? 'https:' + m[1] : undefined; const withCover = cover ? items.map((it) => Object.assign({}, it, { seriesCover: cover })) : items; return { items: withCover, hasNext }; })()"
+  },
+  "pagination": {
+    "mode": "increment",
+    "param": "page_num",
+    "start": 0,
+    "step": 1,
+    "itemsAt": "items",
+    "hasMore": "hasNext",
+    "maxPages": 5
+  },
+  "assert": [
+    {
+      "path": "items",
+      "desc": "tencent vsite_episode_list flattened"
+    }
+  ],
+  "mapping": {
+    "guid": "item_id",
+    "title": "item_params.title",
+    "link": "https://v.qq.com/x/cover/{item_params.cid}/{item_params.vid}.html",
+    "image": "item_params.image_url",
+    "author_avatar": "seriesCover",
+    "description": "item_params.video_subtitle",
+    "pubDate": "item_params.publish_date"
+  },
+  "meta": {
+    "normalizer": "rsshub",
+    "type": "post",
+    "description": "腾讯视频 — 剧集/综艺分集",
+    "topics": [
+      "腾讯视频",
+      "tencent",
+      "v.qq.com",
+      "影视",
+      "综艺",
+      "剧集",
+      "分集",
+      "tv"
+    ],
+    "categories": [
+      "multimedia"
+    ],
+    "facility": {
+      "key": "tencent",
+      "label": "腾讯视频"
+    },
+    "capabilities": [
+      "timeline"
+    ],
+    "key_param": "cid",
+    "priority": 80,
+    "discoverable": true,
+    "cadence_hint_seconds": 86400,
+    "auth": {
+      "type": "none"
+    },
+    "homepage": "v.qq.com",
+    "params_schema": {
+      "cid": {
+        "type": "string",
+        "required": true,
+        "description": "剧集/专辑 cid（取自 v.qq.com/x/cover/{cid}/... 页面 URL）"
+      }
+    },
+    "radar": [
+      "v.qq.com/x/cover/:cid"
+    ]
+  }
+}

package/resources/packages/toubiec/README.md

@@ -0,0 +1,38 @@
+## 说明
+
+nextmusic.toubiec.cn — 另一个第三方网易云解析站（公开入口是 wyapi.toubiec.cn）。和 zuna 不同，
+它**不签名也不加密**：全是明文 JSON POST，所以这几份都是纯 http recipe，compute 只用来拼 body。
+
+## 三份 recipe
+
+| recipe | 干什么 | 谁在用 |
+|---|---|---|
+| `toubiec-search` | 关键词 → 歌曲 | `music-search` 行的并发成员（与 `zuna-search` 并列） |
+| `toubiec-download` | song id → 可播放/可下载地址 | `netease-track` 梯子首档（radar `music.163.com/song`） |
+| `toubiec-playlist` | 歌单 id → 整份曲目名单 | radar `music.163.com/playlist` |
+
+搜索与歌单的 item link 回到 `music.163.com/song?id=`——文件地址在播放/下载那一刻由
+`netease-track` 现解。
+
+## 每个请求都必须带 `ip`
+
+上游校验的是**调用方自报的公网 IP**：少了这个字段一律回 `400 当前非法提交参数`，空串回
+`参数 ip 不能为空`。所以每份 recipe 都先 prefetch 它自己的 `/api/ip`，再由 `sign` 把值拼进 body
+（它的前端就是这么做的）。
+
+实测边界（2026-08-25）：`timestamp` 反过来**根本不校验**（传 `1` 也过），`ip` 只校验非空
+（`0.0.0.0` 也收）——**没有签名可逆**，只是两个明文字段。
+
+## `toubiec-download` 的元数据来自网易云自己
+
+`getSongUrl` 只回地址、码率、大小，不带歌名歌手专辑封面；而下游要拿这些写进音频文件的 ID3
+标签。所以这份 recipe 另外 prefetch 一次 `music.163.com/api/song/detail`（免鉴权，且它本来就是
+这些字段的权威出处），在 decode 里并进结果。
+
+不用上游的 `getSongInfo` 是有原因的：它同样要 `ip`，而 prefetch 之间取不到彼此的结果
+（`http-fetch.ts` 的 prefetch 只吃 `params`），拿不到真实 IP 就只能编一个。
+
+## 专辑：上游自己坏了，没有对应 recipe
+
+`getAlbum` 恒回 `code -462 Failed to fetch album`（四个不同专辑 id 全同），它前端的解析类型
+也只剩「单曲」一档。等它修好再补。

package/resources/packages/toubiec/package.json

@@ -0,0 +1,15 @@
+{
+  "name": "@streamapp/toubiec",
+  "version": "1.0.0",
+  "author": "stream",
+  "keywords": [
+    "stream-recipe"
+  ],
+  "stream": {
+    "type": "recipe",
+    "name": "Toubiec",
+    "tagline": "搜歌 + 解析网易云直链",
+    "facility": "toubiec",
+    "schemaVersion": 1
+  }
+}

package/resources/packages/toubiec/toubiec-download.recipe.json

@@ -0,0 +1,58 @@
+{
+  "version": 1,
+  "kind": "http",
+  "sourceId": "toubiec-download",
+  "request": {
+    "url": "https://nextmusic.toubiec.cn/api/getSongUrl",
+    "method": "POST",
+    "headers": { "content-type": "application/json" }
+  },
+  "compute": {
+    "capabilities": [],
+    "prefetch": [
+      { "as": "ipResp", "request": { "url": "https://nextmusic.toubiec.cn/api/ip", "method": "GET" } },
+      { "as": "detail", "request": { "url": "https://music.163.com/api/song/detail?ids=%5B{id}%5D", "method": "GET" } }
+    ],
+    "sign": "(() => { const ip = (input.pre.ipResp && input.pre.ipResp.data && input.pre.ipResp.data.ip) || '0.0.0.0'; return { body: JSON.stringify({ id: String(input.params.id), level: input.params.level || 'lossless', ip, timestamp: input.now }) }; })()",
+    "decode": "(() => { const d = input.body && input.body.data; if (!d || !d.url) return { song: [] }; const s = ((input.pre.detail && input.pre.detail.songs) || [])[0] || {}; const ext = (String(d.url).split('?')[0].match(/\\.([a-z0-9]+)$/i) || [])[1] || 'mp3'; return { song: [{ url: d.url, format: ext, bitrate: d.br, size: d.size, level: d.level, name: s.name || String(input.params.id), artist: (s.artists || []).map((a) => a.name).filter(Boolean).join('/'), album: (s.album && s.album.name) || '', cover: (s.album && s.album.picUrl) || '' }] }; })()"
+  },
+  "pagination": { "mode": "increment", "param": "page", "start": 1, "step": 1, "itemsAt": "song", "maxPages": 1 },
+  "assert": [{ "path": "song", "desc": "toubiec getSongUrl wrapped as a one-element array (empty = 这首解不出来，让梯子下一档去试)" }],
+  "mapping": {
+    "guid": "url",
+    "title": "name",
+    "author": "artist",
+    "album": "album",
+    "link": "https://music.163.com/song?id={id}",
+    "enclosure_url": "url",
+    "enclosure_type": "format",
+    "format": "format",
+    "image": "cover"
+  },
+  "meta": {
+    "normalizer": "rsshub",
+    "type": "post",
+    "title": "单曲下载",
+    "description": "toubiec 网易云解析 — 单曲下载（song id → 可播放/可下载地址）",
+    "topics": ["toubiec", "网易云", "音乐", "下载", "无损", "music", "netease"],
+    "categories": ["music"],
+    "facility": { "key": "toubiec", "label": "toubiec" },
+    "capabilities": ["timeline"],
+    "key_param": "id",
+    "priority": 80,
+    "discoverable": true,
+    "cadence_hint_seconds": 3600,
+    "auth": { "type": "none" },
+    "homepage": "wyapi.toubiec.cn",
+    "params_schema": {
+      "id": { "type": "string", "required": true, "description": "网易云歌曲 ID" },
+      "level": {
+        "type": "string",
+        "required": false,
+        "default": "lossless",
+        "description": "音质：standard/exhigh/lossless/hires/jyeffect/sky/jymaster（上游给不出时自动降档）"
+      }
+    },
+    "radar": ["music.163.com/song"]
+  }
+}