agentleFS
sitetillersitetiller/ai-visibility-skills/skills/comparison-pages/SKILL.md

comparison-pages

Skill0 starsChanged 2 days agoMITView on GitHub

Plan, write and check honest 'X vs Y' and 'alternatives to X' pages for SaaS: sourced and dated comparison tables, clear disclosure when you are one of the products, no fake or self-written reviews, no self-serving review markup, and measurement with measure-rollback. Based on Google's spam policies and review snippet guidelines and the FTC rule on consumer reviews. Use when the user says 'vs page', 'alternatives page', 'comparison page', 'competitor comparison', 'X alternative', 'compare us to', 'best tools for' page.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 7 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Write the verdicts: who each product fits, and when not to choose yours.

    line 22
  2. Distinctive:

    Add the disclosure at the top (below).

    line 23
  3. Distinctive:

    Every competitor fact has a primary source and a date.

    line 73
  4. Distinctive:

    Do not disparage with unverified claims.

    line 74
  5. Distinctive:

    Do not use competitor trademarks as if endorsed.

    line 75
  6. Distinctive:

    No invented reviews, testimonials, ratings or review counts, by anyone.

    line 76
  7. Distinctive:

    No review markup on comparison or alternatives pages.

    line 77

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Comparison pages that stay true

A comparison page is useful when a buyer can verify it. It earns trust (and avoids policy trouble) by showing what was checked, when, and where the author has a stake. The skill builds pages where every cell in the table has a source and a date, and rejects tactics that look like fake reviews or thin affiliate content.

Inputs

  • The product, the competitor(s), and the buyer question ("X vs Y", "alternatives to X").
  • The user's own pricing and feature facts, and access to each competitor's public pricing, docs and changelog pages.
  • Queries from gsc that the page should serve (see content-brief).

Steps

  1. Collect facts only from primary sources: the competitor's pricing page, documentation, terms, status or changelog pages, and the user's own product. Save each source URL with the access date. Take screenshots of pricing pages the day you read them.
  2. Pick criteria that buyers decide on (price at stated usage, limits, integrations, support, security docs, data export). Take criteria from the queries and from the questions in content-brief, not from what flatters the product.
  3. Build the table (format below). Every cell has a value and a source link. Where the competitor does not publish a fact, write "not published" and the date, never a guess.
  4. Write the verdicts: who each product fits, and when not to choose yours. Include at least one case where the competitor is the better choice, if true. A page where the author always wins is not credible and fails the usefulness test below.
  5. Add the disclosure at the top (below).
  6. Check the page:
    npx sitetiller page https://example.com/vs/acme --json
    npx sitetiller cite https://example.com/vs/acme --json
    npx sitetiller schema https://example.com/vs/acme --json
    page: title, h1, tables (should be at least one), words, canonical. cite: share of sections with numbers (specificShare) and sources (sourcedShare); low values mean unsourced claims. schema: see the markup rules below.
  7. Set a review date and a re-check schedule. Pricing and features change; a stale table is a wrong table.
  8. Measure with measure-rollback: snapshot, publish or edit, compare against similar pages left untouched for 28 days. For a new page, track its queries from gsc by page.

Table format

CriterionYour productCompetitorSource and date
Entry price at stated usage (say the usage)valuevalueURL, checked 2026-11-12
Free plan or trialvaluevalueURLs, dates
Limit that matters (seats, projects, API calls)valuevalueURLs, dates
Feature the buyer asked aboutyes / no / partial, with the detailsamedocs URLs, dates
Data exportvaluevaluedocs URLs
Row labels name the exact plan compared. Put "last checked" and the reviewer's name near the table.

Disclosure

Say it above the fold, in plain words: "This page is published by {Company}, which makes {Product}. We compared the products on {date} using the public sources linked in the table." If affiliate links or paid placements are involved, say so next to the links and mark paid links rel="sponsored" (Google's spam policy treats paid links as acceptable when marked nofollow or sponsored). If the page says it is independent and the company controls it, that is the misrepresentation the FTC rule describes.

What not to do

TacticWhy not
Write "customer reviews" yourself, or have staff, friends or agencies post themThe FTC's rule on consumer reviews and testimonials prohibits fake or false reviews and testimonials; insiders must disclose material relationships
Pay or reward for reviews that require a positive sentimentProhibited by the same FTC rule; Google also says not to include fake or undisclosed incentivised reviews in the page or markup
Present a company-run comparison as independentFTC: misrepresenting a company-controlled site as providing independent reviews is prohibited; the disclosure above avoids it
Hide negative reviews with legal threats or intimidationFTC rule covers review suppression
Copy competitor descriptions or vendor reviews with nothing addedGoogle's spam policy on thin affiliate pages: reusing merchant descriptions and reviews without original value
Generate hundreds of "X vs Y" pages from a template with no original factsGoogle's spam policy on scaled content abuse
Mark up your own scoring of competitors as Review or AggregateRatingSee below
A "comparison calculator" or tool that does not workGoogle lists it under misleading functionality
Quoting a competitor's price without a dateBecomes false silently

Markup

  • Self-serving reviews: Google's review snippet guidelines say that when the reviewed entity controls the reviews about itself, LocalBusiness or Organization pages are ineligible for star results, including reviews shown through a third-party widget.
  • Reviews must refer to a specific item, not a category or a list of items. A comparison or alternatives page reviews several items, so it is not a fit for review markup. Do not add Review or AggregateRating to it.
  • Ratings and reviews must be visible, come from users, and not be aggregated from other websites.
  • Safe markup for these pages: Article, BreadcrumbList, and the Organization/WebSite graph (see schema). A product page of yours can carry SoftwareApplication with real, visible user ratings only.

Output to the user

  1. The page outline: disclosure, short answer, table, per-competitor sections, "who should choose what", sources, review date.
  2. The filled table with every source URL and date, and a list of cells that need a source ("not published" counts as sourced if dated).
  3. The page, cite and schema results and what was fixed.
  4. Measurement plan and the re-check date.

Guardrails

  • Every competitor fact has a primary source and a date; no hearsay, no screenshots of private material.
  • Do not disparage with unverified claims. State what the source shows.
  • Do not use competitor trademarks as if endorsed; name them as the subject.
  • No invented reviews, testimonials, ratings or review counts, by anyone.
  • No review markup on comparison or alternatives pages.
  • This is not legal advice. If the user plans paid placements, affiliate deals or incentivised reviews, tell them to check the FTC guidance and consult a lawyer.
  • Publish only on pages the user owns.

References

  • references/sources.md

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.