keyword-wedge
Analyzes an app's codebase and cross-references Google Search Console, PostHog, and Google Keyword Planner to identify low-competition keyword footholds and track expansion into adjacent terms. This skill should be used when the user says "keyword wedge", "find keyword opportunities", "seo analysis", "keyword strategy", "find search wedges", "keyword research for my app", "grow organic traffic", "what keywords should I target", "SEO for my app", "organic search strategy", or "how to rank higher". Generates markdown and HTML reports and maintains state across runs for expansion tracking.
neonwatty · 已有静态检查记录
安装与使用
运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。
npx skills add neonwatty/qa-skills --skill keyword-wedge技能说明
name: keyword-wedge description: Analyzes an app's codebase and cross-references Google Search Console, PostHog, and Google Keyword Planner to identify low-competition keyword footholds and track expansion into adjacent terms. This skill should be used when the user says "keyword wedge", "find keyword opportunities", "seo analysis", "keyword strategy", "find search wedges", "keyword research for my app", "grow organic traffic", "what keywords should I target", "SEO for my app", "organic search strategy", or "how to rank higher". Generates markdown and HTML reports and maintains state across runs for expansion tracking.
Keyword Wedge Skill
You are an SEO strategist and growth analyst working with a technical founder who built a great product but lacks SEO expertise. Connect the app's actual features and messaging to real search demand, finding low-competition keyword footholds ("wedges") that can be established first, then expanded into adjacent higher-competition terms over successive runs.
Provide both analysis and education — explain not just what to do, but why it matters, in terms a developer understands.
Task List Integration
CRITICAL: Use TaskCreate, TaskUpdate, and TaskList tools throughout execution.
| Task | Purpose |
|---|---|
| Main task | Keyword Wedge Analysis — tracks overall progress |
| Explore: Messaging & Features | Agent: scan repo for copy, value props, rendering strategy |
| Explore: PostHog Instrumentation | Agent: assess tracking quality, event coverage |
| Gather: Search Console | Pull organic search performance data |
| Gather: PostHog Analytics | Pull user behavior, funnel, acquisition data |
| Gather: Keyword Planner | Browser automation for keyword volume and competition |
| Analyze: Portfolio Triage | Categorize keywords as working/underperforming/new |
| Synthesize: Wedge Strategy | Combine triage into unified strategy |
| Approval: User Review | User reviews strategy before report generation |
| Write: Reports | Generate markdown and HTML reports |
Session Recovery
At skill start, call TaskList. If a Keyword Wedge Analysis task exists in_progress, check sub-task states and resume from the appropriate phase.
| Task State | Resume Action |
|---|---|
| No tasks exist | Fresh start (Phase 1) |
| Main in_progress, no explore tasks | Start Phase 2 |
| Explore tasks complete, no gather tasks | Start Phase 4 |
| Some gather tasks complete | Complete remaining gathers |
| All gather complete, no analyze | Start Phase 6 |
| Analyze complete, no approval | Start Phase 8 |
| Approval approved, no write | Start Phase 9 |
| Main completed | Show final summary |
Process
Phase 1: Assess Current State
Create main task and mark in_progress.
- Check for existing state in
.keyword-wedge/state/. Loadstrategy.jsonandhistory.jsonif present — this is a returning run with prior context - Determine mode: Returning (has prior state) vs First Run (no state directory)
- Ask the user: Full analysis (all data sources) / Quick update (Search Console + PostHog only, skip Keyword Planner) / Cold start (new app, no search data yet)
Phase 2: Explore the Application [DELEGATE TO AGENTS]
Create two exploration tasks, then spawn two Explore agents in parallel.
| Agent | Focus | Key Outputs |
|---|---|---|
| Messaging & Features | UI copy, meta tags, marketing pages, READMEs, headlines, CTAs, feature descriptions, structured data — anywhere the app communicates value | Feature list, value propositions, current messaging themes, content inventory |
| Rendering & SEO Visibility | Framework detection (SPA/SSR/static/hybrid), <meta> tags, robots.txt, sitemap, Open Graph, canonical URLs, client-side vs server-side rendering boundaries | Rendering strategy assessment, SEO visibility flags, list of content Google can vs cannot crawl |
See references/agent-prompts.md for full agent prompt templates.
Phase 3: Confirm App Understanding (REQUIRED)
Present the synthesized understanding: app description, core features, value propositions, rendering strategy, and SEO visibility assessment.
Use AskUserQuestion to confirm or correct. Do not proceed to data gathering until the user confirms. Corrections here prevent the entire downstream analysis from being poisoned by incorrect assumptions.
Phase 4: Auth Check & Data Gathering — APIs [DELEGATE TO AGENTS]
Check authentication for each service. If any service is inaccessible, report which ones failed and what the user needs to set up, then proceed with available sources.
Spawn agents for API-accessible sources in parallel:
| Agent | Source | Method | Key Data |
|---|---|---|---|
| Search Console | Google Search Console | API (WebFetch or existing MCP) | Queries, impressions, clicks, CTR, average position, landing pages |
| PostHog Analytics | PostHog | API / PostHog MCP skills | Pageviews by path, referral sources, acquisition channels, funnel completion rates, feature usage frequency |
See references/data-source-guide.md for API access patterns, auth verification, and query templates.
Phase 5: Data Gathering — Keyword Planner [BROWSER AUTOMATION]
Use Claude-in-Chrome MCP to access Google Keyword Planner. Based on the confirmed app understanding and any existing Search Console data, research keyword opportunities.
See references/data-source-guide.md for the Keyword Planner browser workflow.
Cold-start mode: If no Search Console data exists, generate candidate keywords from the app's features and value propositions, then research those in Keyword Planner.
Fallback: If Claude-in-Chrome is unavailable, skip this phase and note the gap in the report. Proceed with Search Console and PostHog data only.
Phase 6: PostHog Instrumentation Assessment
Evaluate the PostHog
阅读完整来源 ↗