All skills
prewsh avatar

/keyword-research

@de3fff0
by prewshprewsh/seo-aeo-engine3 stars
2

An Antigravity AI skill designed to generate high-converting landing pages optimized for Search Engine Optimization (SEO) and Answer Engine Optimization (AEO).

  • 2 files
  • 16.7 KB
  • Updated 2 weeks ago
  • GitHub

Use this Skill: https://skilld.dev/gh/prewsh/seo-aeo-engine/keyword-research

This session only. Nothing lands on disk.

SKILL.md

โ‰ˆ44 tokens always: the name and description. โ‰ˆ3.8k when used: this file.

Keyword Research Skill

Description

Activate this skill when the user wants to research, identify, or prioritise keywords for a topic, niche, or website. Trigger phrases include: "keyword research", "find keywords for", "what should I rank for", "keyword analysis", "best keywords for [topic]", "keyword strategy", "what are people searching for about [topic]", "SEO keywords for [niche]".


Input Format

{
  "topic": "string โ€” the subject to research keywords for",
  "business_type": "string โ€” what the business or site does",
  "target_audience": "string โ€” who the content is for",
  "content_goal": "rank | convert | educate | all",
  "location": "string โ€” optional, country or city for local SEO",
  "intent_focus": "informational | transactional |
                   commercial | navigational | all",
  "competitor_urls": ["string โ€” optional, up to 3 competitor URLs
                       to inform keyword gaps"],
  "user_keywords": ["string โ€” optional keywords the owner already cares about"],
  "site_context": "string โ€” optional audit findings, page list, or product description",
  "research_mode": "browser | api | both | unavailable โ€” default both"
}

Start by asking the owner which keywords, products, audiences, locations, and business outcomes matter. If they do not know, derive provisional candidates from the site, audit findings, competitor language, and observed search results. Do not silently replace owner-supplied priorities with guessed keywords.

Live search-intent research

Use both browser research and available search/keyword APIs when the environment supports them. Prefer APIs for repeatable data and browser inspection for actual result-page intent, related questions, snippets, and current wording. If only one is available, state which source was used. If neither is available, perform semantic research only and mark volume, difficulty, rankings, and SERP observations as unverified.

For each important query, record:

  • the exact query and market/device when known;
  • dominant result intent: learn, compare, evaluate, buy, navigate, or troubleshoot;
  • the problems and questions visible in results, related searches, and snippets;
  • the content formats ranking: landing page, guide, comparison, tool, forum, video, or documentation;
  • the conversion opportunity and the page that should own the query;
  • the date/source and confidence level.

Prioritize problem-related queries that the product can genuinely solve. Search intent outranks attractive but irrelevant volume.


Difficulty & Volume Rubric

Important: This skill estimates difficulty and volume based on semantic analysis and competitive signals. It does not connect to live data APIs. All figures are directional estimates, not verified data. Always validate with a live tool (Ahrefs, SEMrush, Google Keyword Planner) before committing to a full content strategy.

Difficulty Scale

Score Label Meaning
0โ€“20 Easy Low competition, newer or niche keyword
21โ€“40 Moderate Some established content exists, beatable with quality
41โ€“60 Competitive Strong content exists, needs solid authority
61โ€“80 Hard Dominated by high-authority sites
81โ€“100 Very Hard Reserved for the biggest players โ€” avoid early on

Volume Scale (Estimated Monthly Searches)

Label Range
Very High 100,000+
High 10,000โ€“99,999
Medium 1,000โ€“9,999
Low 100โ€“999
Very Low Under 100

CPC Context

CPC (Cost Per Click) signals commercial value even for organic content. Use it to prioritise keywords where advertisers are spending โ€” this means the intent is likely transactional and conversions are valuable.

CPC Range Signal
$0โ€“$1 Low commercial value, good for awareness content
$1โ€“$5 Moderate value, mix of info and commercial intent
$5โ€“$15 High commercial value, strong conversion potential
$15+ Very high value โ€” transactional, competitive space

Output Structure

The skill must always produce output in this exact format:

# Keyword Research Report

**Topic:** [topic]
**Business Type:** [business_type]
**Target Audience:** [target_audience]
**Content Goal:** [content_goal]
**Location:** [location or "Global"]
**Note:** All volume and difficulty figures are estimates.
Validate with a live keyword tool before final decisions.

---

## Seed Keywords

Core terms that define the topic's search territory:

| Seed Keyword         | Est. Volume | Difficulty | CPC Est. | Intent          |
|----------------------|-------------|------------|----------|-----------------|
| [keyword]            | [label]     | [score]    | [$x]     | [intent type]   |
| [keyword]            | [label]     | [score]    | [$x]     | [intent type]   |
| [keyword]            | [label]     | [score]    | [$x]     | [intent type]   |

---

## Priority Keyword Tiers

### ๐ŸŸข Tier 1 โ€” Target First (High Value, Achievable)
Low-to-moderate difficulty. Should anchor your first
wave of content.

| Keyword              | Est. Volume | Difficulty | CPC Est. | Intent        | Content Type to Create  |
|----------------------|-------------|------------|----------|---------------|-------------------------|
| [keyword]            | [label]     | [score]    | [$x]     | [intent]      | [landing page / blog]   |
| [keyword]            | [label]     | [score]    | [$x]     | [intent]      | [content type]          |
| [keyword]            | [label]     | [score]    | [$x]     | [intent]      | [content type]          |

---

### ๐ŸŸก Tier 2 โ€” Build Toward (Medium Difficulty)
Competitive but winnable with topical authority.
Target after Tier 1 content is live.

| Keyword              | Est. Volume | Difficulty | CPC Est. | Intent        | Content Type to Create  |
|----------------------|-------------|------------|----------|---------------|-------------------------|
| [keyword]            | [label]     | [score]    | [$x]     | [intent]      | [content type]          |
| [keyword]            | [label]     | [score]    | [$x]     | [intent]      | [content type]          |

---

### ๐Ÿ”ด Tier 3 โ€” Long-Term Goals (High Difficulty)
Dominated by high-authority sites. Do not target these
until significant topical authority is established.

| Keyword              | Est. Volume | Difficulty | CPC Est. | Intent        | Notes                   |
|----------------------|-------------|------------|----------|---------------|-------------------------|
| [keyword]            | [label]     | [score]    | [$x]     | [intent]      | [why it's hard]         |

---

## Long-Tail Keywords

Lower volume, lower difficulty, higher conversion rate.
These are your fastest wins.

| Long-Tail Keyword                    | Est. Volume | Difficulty | Intent        | Best For              |
|--------------------------------------|-------------|------------|---------------|-----------------------|
| [keyword]                            | [label]     | [score]    | [intent]      | [blog / landing page] |
| [keyword]                            | [label]     | [score]    | [intent]      | [content type]        |
| [keyword]                            | [label]     | [score]    | [intent]      | [content type]        |
| [keyword]                            | [label]     | [score]    | [intent]      | [content type]        |
| [keyword]                            | [label]     | [score]    | [intent]      | [content type]        |

---

## AEO Keywords

Question-based and conversational keywords that AI engines
surface in direct answers, featured snippets, and
People Also Ask boxes.

| AEO Keyword / Question                         | Est. Volume | Intent        | Answer Format to Use   |
|------------------------------------------------|-------------|---------------|------------------------|
| What is [topic]?                               | [label]     | informational | Definition sentence    |
| How does [topic] work?                         | [label]     | informational | Numbered steps         |
| What is the best [topic] for [audience]?       | [label]     | commercial    | Comparison table       |
| How much does [topic] cost?                    | [label]     | transactional | Direct number + range  |
| [topic] vs [alternative] โ€” which is better?   | [label]     | commercial    | Comparison block       |
| Is [topic] worth it?                           | [label]     | commercial    | FAQ answer block       |
| [Common question your audience asks]           | [label]     | [intent]      | [format]               |
| [Common question your audience asks]           | [label]     | [intent]      | [format]               |

---

## LSI and Semantic Keywords

Related terms that reinforce topical authority.
Use these naturally throughout content โ€”
do not force them as primary targets.

- [LSI term 1]
- [LSI term 2]
- [LSI term 3]
- [LSI term 4]
- [LSI term 5]
- [LSI term 6]

---

## Keywords to Avoid

These keywords are flagged as high-risk or low-ROI
for this topic and audience:

| Keyword              | Reason to Avoid                                        |
|----------------------|--------------------------------------------------------|
| [keyword]            | Wrong intent โ€” attracts audience not looking to convert|
| [keyword]            | Too broad โ€” dominated by global brands                 |
| [keyword]            | Cannibalises [other keyword] โ€” similar enough to split traffic |
| [keyword]            | Seasonal โ€” only relevant [month/period]                |

---

## Cannibalization Check

Keywords with enough overlap that two pages targeting
them would compete against each other:

| Keyword A            | Keyword B            | Risk Level | Resolution                          |
|----------------------|----------------------|------------|-------------------------------------|
| [keyword]            | [keyword]            | High/Med   | [which page should own which term]  |

If none detected: โœ… No cannibalization risks found.

---

## Keyword โ†’ Content Map

Recommended content to create based on this research,
in production order:

| Priority | Content Title                    | Target Keyword    | Content Type     | Goal              |
|----------|----------------------------------|-------------------|------------------|-------------------|
| 1        | [title]                          | [keyword]         | Landing Page     | Convert           |
| 2        | [title]                          | [keyword]         | Pillar Blog Post | Rank + Authority  |
| 3        | [title]                          | [keyword]         | Cluster Article  | Topical Depth     |
| 4        | [title]                          | [keyword]         | Cluster Article  | AEO / Snippet     |
| 5        | [title]                          | [keyword]         | FAQ Page         | AEO / Long-tail   |

---

## Quick Wins Summary

The 3 keywords to act on immediately:

1. **[keyword]** โ€” [one sentence on why this is the best
   first target]
2. **[keyword]** โ€” [one sentence on why]
3. **[keyword]** โ€” [one sentence on why]

Foundational Content Plan

Recommend between 5 and 10 foundational pages, defaulting to 10 when the topic has enough distinct intent. Include the landing page when it is a target page, then assign each page a problem, query, intent, conversion path, supporting evidence, and internal-link targets.

# Page / Article Primary Query Problem Solved Intent Conversion Path Evidence
1 [title] [query] [problem] [intent] [CTA/product page] [source/confidence]

Do not fill the table with near-duplicates to reach ten. If fewer than five distinct intents are defensible, explain why and ask whether the owner prefers a smaller foundation or broader adjacent topics.

Search-Intent Evidence

Query Market / Date Dominant SERP Intent Ranking Formats User Problem Source Confidence
[query] [market/date] [intent] [formats] [problem] [browser/API/semantic] High/Med/Low

Execution Steps

  1. Parse topic, business_type, site_context, and any user_keywords together to define the full semantic territory โ€” not just the obvious head term but adjacent terms the audience uses
  2. Generate seed keywords โ€” 3โ€“5 core terms that anchor the topic
  3. Expand each seed keyword into long-tail variations using question formats, modifier patterns ("best", "how to", "for [audience]", "vs"), and location modifiers if location is provided
  4. Generate AEO keyword list โ€” convert the most common audience questions into direct-answer keyword targets
  5. Generate LSI terms โ€” identify semantically related vocabulary that reinforces topical authority without targeting directly
  6. If live browser or API research is available, collect current search-intent evidence before estimating difficulty. Base any remaining estimate on keyword length, specificity, and likely competitive landscape
  7. Estimate volume using the volume scale โ€” shorter, broader terms get higher volume labels; longer, more specific terms get lower
  8. Assign intent category to every keyword
  9. Sort all keywords into Tier 1, 2, and 3 based on difficulty and estimated volume
  10. Run cannibalization check โ€” flag any two keywords that are similar enough to split traffic if targeted on separate pages
  11. Build Keywords to Avoid list โ€” flag wrong-intent, too-broad, cannibalistic, and seasonal keywords
  12. Build the Keyword โ†’ Content Map โ€” recommend the landing page, 5โ€“10 foundational pages (default 10), supporting articles, and production order. Every page must own a distinct intent.
  13. Mark which keyword choices require owner confirmation before content production and which are safe provisional fallbacks.
  14. Execute scripts/score_keywords.py to verify difficulty estimates programmatically โ€” if unavailable, proceed manually and note "Script verification skipped" in output

Script Reference

Script: scripts/score_keywords.py Invocation: python scripts/score_keywords.py --topic "[topic]" --keywords "[comma_separated_list]" Outputs: difficulty scores, intent classification, long-tail expansion suggestions Fallback: if script unavailable, complete all analysis manually using the rubrics above and note "Script verification skipped" in the output


Connected Skills

  • Receives output from: orchestrator workflow input
  • Feeds output to: content-cluster, landing-page-writer, blog-writer, meta-description-generator

Source: SKILL.md on GitHub

No third-party reports yet.

Signed by skilld at de3fff0. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 2 weeks ago.

Activeupdated 2 weeks ago

README badge

README badge for prewsh/seo-aeo-engine/keyword-research