據 AlphaMoat 統計,截至 2026-09,Scholar 累計下載 4,734 次、獲 6 個星標,在全部 5,078 個 AI 技能中排名第 1,996 位,在「知識管理」分類(442 個)中排名第 187 位。
| 平台 | 累計安裝 | 累計下載 | 星標 | 評論 |
|---|---|---|---|---|
| ClawHub ↗ | 140 | 4369 | 3 | - |
| SkillHub ↗ | 526 | 4734 | 6 | — |
各平台的計數口徑不同,統計時間不一致。
該技能提供 Google Scholar 學術文獻檢索的實用規則,涵蓋查詢構造、時間篩選、結果解讀、全文獲取、引用分析、提醒追蹤,以及侷限性認知與來源評估。它宣告支援多種作業系統,是一個純檢索指南類技能,不依賴二進位制或金鑰。 這是手動使用 Scholar 的檢索指南,不會自動搜尋全文或判斷品質。引用前核對論文與發表來源。
資料釋出於 2026-10-27
以下是根據技能文件生成的示例,不代表實際執行結果。
2026-09 起統計方法調整:同名技能按作者分別統計,並覆蓋全部技能。圖中虛線為調整前的月份,與之後的資料不直接比較,也不計算跨越調整的增長。
| 月份 | 累計下載 | 環比 | 星標 |
|---|---|---|---|
| 2026-09 | 4734 | — | 6 |
| 2026-08 | 4166 | +6.3% | 6 |
| 2026-07 | 3920 | +10.4% | 6 |
| 2026-06 | 3552 | +5.8% | 3 |
| 2026-05 | 3356 | — | 5 |
| 2026-03 | 2573 | — | 3 |
以下是作者釋出的原始檔案,按純文字顯示,本站未做修改或稽核;其中的連結不可點選,命令請核實後再執行。抓取於 2026-10-23。
---
name: Scholar
description: Search academic literature with Google Scholar using effective queries, citations, and filters.
metadata: {"clawdbot":{"emoji":"🎓","os":["linux","darwin","win32"]}}
---
# Google Scholar Research Rules
## Query Construction
- Exact phrases in quotes: "machine learning" finds the phrase, not separate words- Exclude terms with minus: neural networks -deep removes deep learning results - OR for alternatives: "global warming" OR "climate change" — must be uppercase - author: operator for specific researchers: author:"Y LeCun" - source: for specific journals: source:"Nature" quantum computing - intitle: forces term in title: intitle:transformer attention ## Time Filters - Custom range critical for recent research — default shows highly cited old papers first - "Since 2023" for cutting-edge but less vetted work - "Since 2020" balances recency and citation accumulation - Sort by date for newest, sort by relevance for most cited — toggle based on need ## Understanding Results - Citation count indicates influence, not quality — popular isn't always right - "Cited by X" link shows who built on this work — follow research forward - "Related articles" finds similar work — useful after finding one good paper - Versions link shows preprints and alternate copies — often free when journal isn't - Quotation marks in snippet show exact matches — verify relevance before clicking ## Finding Full Text - Check "All versions" for free copies — preprints, institutional repositories - Add filetype:pdf to find direct PDF links - Unpaywall browser extension auto-finds legal free versions - Request from authors directly — most will share, academia.edu and ResearchGate common - Institutional access through library proxy — VPN to university network ## Citation Analysis - h-index visible on author profiles — measures productivity and impact - "Cited by" sorted by relevance shows most influential citations - Citation count alone misleads — review papers cite heavily but add less - Self-citations inflate metrics — check who's actually citing - Recent citations more meaningful for living research — old papers cited by habit ## Alerts and Tracking - Create alerts for search terms — email when new papers match - Follow researchers via profiles — notifications for new publications - "Cite" button exports to reference managers — BibTeX, EndNote, RefMan - Save to library requires Google account — organizes papers for later ## Limitations to Know - Coverage biased toward English and STEM — humanities and non-English underrepresented - No quality filter — predatory journals appear alongside legitimate ones - Older papers may not be indexed — pre-digital era incomplete - Cannot search full text — only titles, abstracts, and metadata - Ranking algorithm opaque — not always clear why results ordered as shown ## Evaluating Sources - Check journal reputation before trusting — impact factor as rough guide - Preprints not peer-reviewed — valuable for speed, treat as preliminary - Conference papers vary by field — top-tier in CS, lower status in medicine - Retracted papers may still appear — verify paper status before citing - Thesis and dissertations are gray literature — use cautiously ## Search Strategies - Start broad, narrow with filters — missing papers worse than sorting many - Combine Scholar with domain databases — PubMed for medicine, IEEE for engineering - Snowball: find one good paper, explore its citations and references - Check review papers for comprehensive coverage — synthesize existing knowledge - Multiple queries with synonyms — terminology varies across disciplines ## Common Mistakes - Trusting citation count as quality measure - Ignoring publication date — fields evolve, old papers may be outdated - Stopping at first page of results — good papers may be buried - Not checking for retractions or corrections - Citing papers based only on abstract — always read methodology - Missing preprint versions that are freely available
把長文字、文件與網頁內容壓縮成要點和關鍵詞的摘要工具。
透過 IMA OpenAPI 管理筆記與知識庫內容。
規定如何為網頁請求選擇合適工具的必讀指引。
呼叫百度 AI 搜尋介面做網頁檢索的技能。
用 uvx 直接呼叫 markitdown,把各類檔案轉成 Markdown。
面向 OpenClaw 的預設網頁搜尋,即時檢索多個引擎。
用 DuckDuckGo 介面做網頁、新聞、圖片與影片檢索的技能。
檢索並發現可用 OpenClaw 技能的技能。
將使用者明確糾正與自我檢視整理成分層記憶,並依使用情況升級或封存。
在保留公式、格式和活頁簿結構的前提下建立、檢查或編輯 Excel 檔案。
處理 Word 與 DOCX 文件的作業指南,著重維持樣式、編號與版面。
把資料庫、表格與匯出的原始資料變成可決策結論的資料分析技能。
協助編輯 PowerPoint 簡報,並在交付前檢查版面是否正常呈現。
資料月份:2026-09 · 下載/星標/安裝數聚合自公開技能註冊庫(ClawHub、SkillHub),按月更新。口徑說明見資料方法。