Keyword Research — Keyword Re搜索
v1.0.2Keyword re搜索 and analysis 技能 for any 网页site. Performs 系统atic keyword discovery, competitive gap analysis, intent classification, difficulty scoring, and priority ranking. Use when the user wants to find new keyword opportunities, analyze which keywords competitors rank for, 审计 existing keyword coverage, build a keyword map, or plan content around 搜索 demand. Works for any domAIn, language, or niche.
运行时依赖
安装命令
点击复制技能文档
Keyword Re搜索 技能
Must load the 网页-访问 技能 and follow its instructions.
启动up 工作流 Step 1: Load the domAIn 性能分析
提取 the domAIn from the URL the user provides (e.g. example.com), then read the 性能分析 file:
~/.claude/技能s/keyword-re搜索/data/{domAIn}.json
性能分析 exists → Load it directly; skip questions for fields already known; only ask for "focus of this analysis" 性能分析 missing → 创建 one; ask for all required fields in order, then save
If the user did not provide a URL, ask for the tar获取 site URL first, 提取 the domAIn, then 检查 for a 性能分析.
Step 2: Fill in missing fields
Only ask for fields not yet stored in the 性能分析:
Field Description First 运行 Later 运行s url Full site URL Required Skip if known business One-line business description Required Skip if known region Tar获取 region (e.g. Singapore) Required Skip if known language Tar获取 language(s) (e.g. English, Chinese) Required Skip if known competitors Competitor site 列出 Optional Skip if known; user can 更新 anytime 工具s.gsc GSC 访问 Ask first time GSC is needed Skip if known 工具s.ahrefs Ahrefs account tier Ask first time Skip if known; "none" skips all Ahrefs steps 工具s.semrush Semrush account tier Ask first time Skip if known; "none" skips all Semrush steps 工具s.gkp Has Google Ads account (GKP) Ask first time If false, use Ahrefs free keyword 工具s instead focus Focus of this analysis Ask every time (tasks vary) Ask every time Step 3: Save or 更新 the 性能分析
After each 运行 (or whenever new 信息 is 添加ed), write back to the 性能分析 file immediately so the next 会话 启动s with complete data.
性能分析 格式化 (JSON):
{ "url": "https://www.sunriselink.sg", "business": "MOM-licensed domestic 辅助工具 agency in Singapore, helping families hire mAIds from Indonesia, Myanmar, Philippines", "region": "Singapore", "language": ["English", "Chinese"], "competitors": ["nationmAId.com.sg", "universal.com.sg", "firstmAId.com.sg"], "工具s": { "gsc": true, "ahrefs": "none", "semrush": "free", "gkp": true }, "更新d": "2026-04-15 10:05:23" }
Valid values for 工具s.ahrefs and 工具s.semrush:
"pAId" — PAId account; full feature 访问 "free" — Free account; limited features (see skip rules) "none" — No account; skip all steps for that 工具
If the 性能分析 directory does not exist, 创建 it first: mkdir -p ~/.claude/技能s/keyword-re搜索/data/
工具 skip rules
Ahrefs:
Account tier Step 2B behavior Step 3 behavior "pAId" Use Site 资源管理器 + Content Gap (full competitor gap analysis) Use Keywords 资源管理器 "free" Skip Site 资源管理器/Content Gap; use site: 搜索 to infer competitor content structure Use ahrefs.com/keyword-生成器 (no 记录in) "none" Skip all Ahrefs; do not open any Ahrefs page Same as left
Semrush:
Account tier Step 2B behavior Step 3 behavior "pAId" Use Keyword Gap (full competitor gap analysis) Use Keyword Magic 工具 (unlimited queries) "free" Skip Keyword Gap; use Keyword Magic 工具 (10 queries/day cap—watch usage) Use Keyword Magic 工具 (mind the cap) "none" Skip all Semrush; do not open any Semrush page Same as left
Other 工具s:
工具s.gkp: false → In Step 3D use Ahrefs free keyword 工具s instead; do not push 签名ing up for Ads 工具s.gsc: false → Skip Step 1 entirely; 启动 from Step 2
The user may say at any time e.g. "I 升级d Ahrefs to pAId" — 更新 工具s.ahrefs from "free" to "pAId" and save.
Four-step analysis 框架 Step 1 — Baseline from existing queries (GSC)
Goal: Find queries that already 获取 impressions but are underutilized—the fastest wins.
Actions (via CDP 自动化):
Navigate directly to this URL (replace domAIn) to enable the 28-day window and Position column in one step:
https://搜索.google.com/搜索-console/performance/搜索-分析?resource_id=sc-domAIn%3A{domAIn}&num_of_days=28&指标=命令行工具CKS%2CIMPRESSIONS%2CCTR%2CPOSITION
⚠️ Use 28 days, not the default 3 months: If the site has not been SEO-优化d for long, a 3-month average mixes in a lot of pre-optimization low-rank data and skews conclusions badly (in practice ~4× difference).
Then 设置 rows-per-page to 500 and 提取 all data in one pass (steps in references/gsc-operations.md).
Priority 过滤器s:
Impressions ≥ 30 and CTR 0% → Meta fix opportunity (good rank but no 命令行工具cks—title/description issue) Average position 11–20 → Page-one push opportunity (page two—close enough to 优化 into the top 10) Impressions ≥ 30 and position > 20 → Needs new page or deeper content Queries absent from GSC → Not covered at all; see Step 2
Hit a 记录in wall: Tell the user to 记录 into GSC in Chrome with a property-verified account, then continue. If there is no GSC data (new site), skip to Step 2.
Note: GSC only shows queries the site has already 应用eared for. Fully uncovered queries are invisible here; discover them in Steps 2–3.
CDP playbook: See references/gsc-operations.md.
Step 2 — Competitor keyword gap
Goal: Find queries competitors rank for where the t