Stats
Actions
Tags
From novel-looker
給一個小說網站,分析其 DOM 結構並產出可被 novel-looker CLI 直接匯入的 JSON 書源。觸發詞:「解析這個小說網站」「幫我寫書源」「新增書源」「parse novel site」。輸入需要至少一個搜尋頁 URL 或書籍詳情頁 URL。
How this skill is triggered — by the user, by Claude, or both
Slash command
/novel-looker:parse-novel-siteThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
把一個小說網站變成一份能被 `novel-looker` 抓的 JSON 書源。
把一個小說網站變成一份能被 novel-looker 抓的 JSON 書源。
跟使用者確認以下至少其中一個(越多越準):
{{key}})如果只給了網域,請主動搜尋一個關鍵字、或抓首頁找熱門書連結作為樣本。
對每個樣本 URL:
curl -sL -A "Mozilla/5.0" <url> 取 HTMLli.book),以及子節點的 title / author / book_urltag > tag),避開 framework 隨機 hash<css_selector>[@<accessor>][##<regex>##<replacement>]
accessor:text (預設) / html / outerHtml / <attr 名> (如 href、src)|| 連接 fallback:.title || h1& 表「當前節點本身」(例如 chapterName: "&@text"、chapterUrl: "&@href")寫入 book-sources/<site_name>.json,欄位完整範本:
{
"bookSourceUrl": "https://example.com",
"bookSourceName": "示例小說網",
"bookSourceGroup": "中文/玄幻",
"enabled": true,
"bookUrlPattern": "https?://example\\.com/book/\\d+",
"ruleSearch": {
"url": "https://example.com/search?q={{key}}",
"bookList": "...",
"name": "...",
"author": "...",
"bookUrl": "...@href"
},
"ruleBookInfo": {
"name": "...",
"author": "...",
"intro": "...",
"coverUrl": "...@src",
"tocUrl": "...@href"
},
"ruleToc": {
"chapterList": "...",
"chapterName": "&@text",
"chapterUrl": "&@href"
},
"ruleContent": {
"content": "..."
}
}
cargo run -- source import book-sources/<site_name>.json
cargo run -- search "<keyword>" # 確認 ruleSearch
cargo run -- add --source <bookSourceUrl> <book_url> # 確認 ruleBookInfo
cargo run -- sync <novel_id> # 確認 ruleToc
cargo run -- read <novel_id> 0 # 確認 ruleContent
任一步失敗 → 回頭重抓對應頁面、修選擇器、再跑。不要在沒驗證的情況下宣稱完成。
TODO: brittle selectorCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub tlogben/novelreader