Anycap Ai Tool Seo

AnyCap AI Tool SEO

Read this entire file before starting. It covers the full SEO planning and audit workflow for AI tool websites.

Plan and audit SEO for AI tool websites. Focus on product-led SEO where each page should help a searcher complete a task, evaluate options, or enter the product with high intent.

Map intent. Shape pages. Add evidence. Protect quality.

Before You Start

Read all three reference files before taking any action. They contain the working framework, page planning rules, and quality guardrails.

1. Read this file for the overview and process 2. Read the references in this order:

• framework.md
• page-planning.md
• guardrails.md

3. Then begin the workflow below

For detailed command syntax, flags, and output parsing, refer to the anycap-cli skill.

Prerequisites

• anycap CLI installed and authenticated (anycap status to verify)
• A local workspace for notes, saved search results, and page briefs

When to Use This Skill

• AI tool website SEO planning
• SaaS or product-led SEO audits
• Search intent to page type mapping
• Page brief creation for tool, comparison, alternatives, pricing, and tutorial pages
• Technical SEO prioritization for new or growing tool websites
• Citation, directory, or backlink planning
• Programmatic SEO evaluation and rollout gating

SEO Planning Process

graph LR
    A[Understand Site and ICP] --> B[Inspect Live SERP]
    B --> C[Choose Page Type]
    C --> D[Define Information Gain]
    D --> E[Check Technical and Trust Signals]
    E --> F[Prioritize and Deliver]

Work through the steps below in order. Skip only when you already have high-confidence answers.

1. Understand the site and the searcher

• Crawl the homepage, pricing page, docs/help center, and one representative tool or feature page.
• Infer:
• the core job-to-be-done
• primary user, buyer, and learner
• target geography and language
• primary conversion event
• Use the 6-field ICP template in framework.md.
• If the user only provides a URL, infer first and ask follow-up questions only when the missing context would materially change the plan.

2. Inspect the live SERP before recommending content

• Do not decide the content format from the keyword alone.
• Use anycap search --query "<keyword>" --no-crawl --max-results 10 to classify dominant page types, recurring modules, and SERP mix.
• Treat search intent as a page-shape constraint, not just a label.
• If the SERP is mixed, decide whether the keyword deserves one page with a dominant intent or multiple pages.
• Use page-planning.md for page type mapping and module requirements.

3. Evaluate page viability through four lenses

• Review every page or keyword cluster through:
• Search-Fit Product: Can the user complete the task on the page or move naturally into the product?
• Information Gain: What first-hand evidence, screenshots, data, tests, workflows, or examples make this page stronger than the current SERP?
• Technical Readiness: Can search systems crawl, index, render, and understand the page?
• Trust Distribution: What internal links, external mentions, backlinks, directory placements, or trust blocks support the page?
• Use "Last-Click" only as a user-satisfaction heuristic. Do not present it as an official Google ranking factor. See guardrails.md.

4. Prioritize the plan

• Default prioritization:

1. money pages with clear transactional or commercial intent 2. comparison, alternatives, and pricing pages 3. tutorials that support discovery, trust, and internal linking 4. pSEO only after a small set of hard pages proves quality and indexation

• Separate:
• high-confidence rules
• practitioner heuristics
• assumptions that still need validation

5. Deliver concrete outputs

• Default deliverables:
• ICP summary
• keyword cluster -> intent -> page type table
• priority page briefs
• technical baseline checklist
• citations / backlinks backlog
• pSEO go / no-go decision with safeguards
• 30 / 60 / 90 day sequencing

Human-in-the-Loop

This skill benefits from light user input up front, but should otherwise run autonomously.

• Ask for the site URL, target market, and conversion goal if they are not clear.
• If the user already gave a concrete site or keyword set, do not over-clarify.
• Once direction is clear, continue through SERP inspection, planning, and prioritization without repeated interruptions.

Core Principles

Define the searcher before the keyword. A keyword only makes sense once you know who is searching, why, and what they need to finish.

Inspect the live SERP before choosing a page type. Do not recommend a tutorial, comparison, or tool page until you know what the current SERP rewards.

Treat page type as intent execution. A page is not just content; it is the shape through which the intent gets fulfilled.

Require evidence, not generic prose. Information gain should come from screenshots, examples, data, workflows, benchmarks, or other concrete proof.

Protect quality before scale. Do not recommend pSEO until high-value sample pages have proven useful, indexable, and maintainable.

Separate rules from heuristics. Be explicit about what is a hard constraint versus what is a useful but situational tactic.

Quick Reference

# Inspect a target site
anycap crawl https://example.com

# Inspect the SERP shape for a keyword
anycap search --query "best ai headshot generator" --no-crawl --max-results 10

# Ask for a grounded summary when the SERP is unclear
anycap search --prompt "What page types dominate the SERP for 'best ai headshot generator' and which content blocks recur?"

Save important search and crawl outputs locally when the task is large or when you expect to revisit evidence.

Guardrails

• Do not promise rankings.
• Do not recommend doorway pages, spun pages, or low-value mass AI pages.
• Do not copy a competitor layout blindly; infer the SERP expectation, then add original evidence.
• Treat numeric thresholds, directory filters, and DR/DA cutoffs as heuristics rather than fixed rules.
• Flag compliance risk when recommending paid placements, sponsored links, or directory submissions.
• When suggesting pSEO, require unique fields, update mechanisms, quality checks, and pruning rules.

Resources

• framework.md -- core model, support levels, and default prioritization
• page-planning.md -- intent mapping, page modules, and page brief outputs
• guardrails.md -- quality boundaries, safety checks, and pSEO gating

Framework

Use this file to keep the stable, reusable parts of the method while separating hard constraints from heuristics.

Core Model

Treat AI tool SEO as five linked systems:

1. ICP and task definition 2. Search-Fit Product 3. Information Gain 4. Technical visibility 5. Trust distribution

1. ICP and task definition

Define the searcher before the keyword. Use this 6-field template:

If the user did not provide ICP details, infer them from the site, but mark the inference clearly.

2. Search-Fit Product

Ask whether the page helps the user finish the job on-page or move naturally into the product.

Common mapping:

• transactional intent -> tool page, feature page, demo page, pricing page
• commercial intent -> comparison page, alternatives page, buyer guide
• informational intent -> tutorial, workflow guide, glossary page, template explainer
• navigational intent -> homepage, login page, docs page, API page, brand page

3. Information Gain

Prefer evidence that cannot be cheaply cloned:

• original screenshots
• tested outputs
• benchmarks
• before / after examples
• real workflows
• product constraints and trade-offs
• proprietary data or carefully structured data

Treat information gain as a content standard, not as a magic ranking formula.

4. Technical visibility

Technical work should make pages crawlable, indexable, renderable, and understandable.

Minimum baseline:

• robots.txt
• sitemap.xml with useful lastmod values
• canonical handling
• correct status codes and redirects
• clear titles and snippets
• image alt text
• appropriate structured data
• acceptable page experience and Core Web Vitals

5. Trust distribution

Do not reduce this to "get backlinks."

• internal links from hubs to money pages
• mentions or placements in directories, review sites, communities, and resource pages
• sponsored exposure when it is genuinely useful and compliant
• on-page trust blocks: customer logos, reviews, editor picks, media mentions

Support Levels

Use this table to control certainty and tone in the final plan:

Default Prioritization

Default order:

1. pages closest to conversion 2. pages that win commercial investigation 3. tutorial content that supports trust and internal linking 4. pSEO only after a small set of hard pages has proven itself

Guardrails

Use this file to control boundaries and avoid presenting heuristics as hard rules.

Public guidance takes priority

When a practical tactic conflicts with public search guidance, let the tactic yield to the guidance.

Priority directions:

• content should be useful to people first
• pages must be crawlable, indexable, renderable, and understandable
• do not use low-value content at scale to manipulate rankings
• generative AI is allowed, but not as pure volume without added value
• page experience matters, but it should not be reduced to a single score game

What is heuristic, not law

These can be useful, but should not be written as universal truth:

• "Last-Click" as a central SEO explanation
• fixed thresholds for directory selection
• a single formula such as "precise traffic > authority = relevance"
• defaulting to deprioritize all informational content in a zero-click era

Treat them as hypotheses to validate with the live SERP, the site state, and the business goal.

What not to claim

• Do not say dwell time, bounce rate, or Chrome data are officially acknowledged ranking factors.
• Do not say directory submissions guarantee ranking gains.
• Do not say AI-generated content is inherently bad.
• Do not say more pages automatically means stronger SEO.

Safe boundaries for links and citations

• Prefer relevant, real, editorial, or genuinely useful mentions.
• Flag compliance risk when paid placements, sponsored links, or paid directory submissions are involved.
• Do not recommend link schemes, bulk link exchanges, or low-quality directory volume plays.
• Treat directory submissions or citations as a focused backlog, not a numbers game.

pSEO go / no-go

Green lights

• unique structured data
• a clear user task on every page
• real variation in page fields
• value blocks in the template, not just keyword swaps
• an update mechanism
• a pruning or noindex plan
• sample pages already meet an acceptable quality bar

Red flags

• only the keyword, city, or brand name changes
• the page does not solve the query and only funnels users elsewhere
• obvious scraping, rewriting, or synonym-swapping
• no unique evidence
• no update path
• no realistic way to clean up weak pages later

If red flags dominate, do not recommend launching pSEO yet.

Page Planning

Use this file to turn keywords and SERP patterns into actual pages.

Intent to page type mapping

Do not force clearly conflicting intents into one page just because the keyword stem is similar.

Core modules by page type

Tool or feature page

• clear outcome statement
• input / output preview or demo
• screenshots or sample outputs
• use cases
• trust proof
• FAQ
• primary CTA
• relevant schema

Comparison or vs page

• comparison table
• methodology or evaluation lens
• strengths and trade-offs
• "best for" guidance
• migration or switching notes
• FAQ
• CTA

Alternatives page

• why users switch
• competitor comparison blocks
• pricing and feature deltas
• scenarios where each option is a better fit
• reviews or proof
• CTA

Pricing page

• clear plans
• feature matrix
• usage limits
• objection handling
• trust and billing FAQ
• CTA

Tutorial page

• quick answer near the top
• steps
• screenshots
• sample outputs
• common mistakes and edge cases
• related tools or templates
• CTA only after usefulness is established

SERP reading checklist

For the top 10 results, record at least:

• dominant page type
• dominant intent
• whether the SERP is mixed
• recurring modules
• common proof patterns
• freshness expectations
• whether community or UGC results appear

If the SERP is clearly mixed, decide whether you need:

• one page with a dominant intent plus limited supporting modules, or
• two or more separate pages

Recommended output format

Prefer a table like this:

One-page brief template

For every priority page, produce:

• target keyword or keyword cluster
• ICP slice
• dominant intent
• recommended URL slug
• page type
• must-have modules
• evidence gaps to fill
• internal links needed
• schema candidate
• conversion goal