Free SEO skills for Claude, ChatGPT, and Cursor

Install a skill and it uses budgetSEO for keyword, competitor, and domain data. The page and the files are free. A fresh lookup spends the same credits as the app, and the skill asks before that call. Reads of data already in the account stay free.

Connector setup stays on the MCP page. A logo connects that client and hands it the skill. None of these clients can install a skill file in one click.

Machine docs: https://www.budgetseo.ai/skills.md · skills index · llms.txt · Updated .

Standing skills

Start with the growth stage, then the smallest skill that matches the client.

  • Growth stage

    You get

    A one-sentence call: this client needs new pages, or this client needs fixes to pages that already get clicks.

    Why use it

    That call comes before any task list. A new site and a site that already ranks should not get the same plan.

    Ask when

    • choose a growth stage
    • new client SEO plan
    • build pages or fix pages
    • thin site or mature site
  • Audit router

    You get

    A short ordered list of which skills to run, and what a person must confirm before anything is published.

    Why use it

    A request to audit the client is too vague to start editing. This picks the smallest check that matches the complaint.

    Ask when

    • audit this client
    • SEO audit
    • what is wrong with this site
    • run an audit
  • Keywords that bring customers

    You get

    A shortlist of queries a buyer would type, each tied to one URL the site should use or already has.

    Why use it

    The useful result is the queries that can bring customers, with one page per topic. A long keyword dump is not a plan.

    Ask when

    • find keywords that bring customers
    • keyword gaps
    • build topical authority
    • queries we should rank for
  • Title tests

    You get

    A small set of replacement titles for pages that already rank and get fewer clicks than they should.

    Why use it

    A clearer title can earn the click without a new page. The test stays small so you can see whether clicks actually move.

    Ask when

    • improve titles
    • click-through rate
    • pages that rank but get few clicks
    • snippet test
  • Refresh stale pages

    You get

    The stale facts to replace, a source for each one, and a date change only when the page itself changed.

    Why use it

    Changing the date on an unchanged page is not a refresh. This finds the claims that are actually out of date.

    Ask when

    • refresh old content
    • outdated pages
    • update the date
    • decaying pages
  • Title, meta, and H1

    You get

    The title, meta description, or H1 that no longer matches the page, and the smallest edit that fixes it.

    Why use it

    After a rewrite, the snippet can promise something the page no longer says. This catches that before it ships.

    Ask when

    • title no longer matches
    • meta description drift
    • H1 mismatch
    • check the snippet after an edit
  • Overlapping URLs

    You get

    Which URL should stay, and which overlapping URL should merge, redirect, or narrow its topic.

    Why use it

    Two pages that answer the same buyer question split the clicks. One question should have one URL.

    Ask when

    • cannibalization
    • overlapping URLs
    • two pages for one query
    • programmatic overlap
  • Pages search engines can reach

    You get

    The important URLs search engines cannot reach, and the one change that removes each block.

    Why use it

    A strong page does nothing if noindex, the sitemap, robots.txt, or a bad canonical keeps it out of search.

    Ask when

    • pages not indexed
    • noindex
    • sitemap missing
    • robots.txt
    • canonical
    • soft 404
    • make important pages accessible
  • On-page consistency

    You get

    The tags, headings, or structured data that contradict the visible page, plus JSON-LD that matches that page.

    Why use it

    A missing optional tag is not the problem. A contradiction between the tags and what a person can read is.

    Ask when

    • technical SEO
    • heading hierarchy
    • structured data
    • canonical mismatch
    • meta tags
  • Who is behind the page

    You get

    The pages missing a real byline or a visible update process, and the fact the client has to supply.

    Why use it

    Readers need to see who is behind the page. This skill will not invent an expert to fill the gap.

    Ask when

    • who wrote this
    • author byline
    • about page
    • invented expert
    • update process
  • Internal links people will use

    You get

    Rows you can hand to the CMS: the source page, the destination, the anchor, and the sentence to change.

    Why use it

    Links already on the site move people before any new outreach. Run this before asking other sites for links.

    Ask when

    • add internal links
    • orphan pages
    • link from an existing page
    • internal links before outreach
  • Link quality

    You get

    Which links matter, which to ask removed, and a disavow list only when each URL has a stated reason.

    Why use it

    A spam score is not a plan. Asking for a link to be removed comes before a disavow file, and a mass disavow does not.

    Ask when

    • toxic links
    • disavow
    • link quality
    • referring domains
    • backlink profile
  • One source per figure

    You get

    Every figure in a draft paired with one source the client can open, or the figure removed.

    Why use it

    Unsourced numbers make a draft sound generic. If the source cannot be checked, the number does not stay.

    Ask when

    • cite a source
    • check a statistic
    • information gain
    • is this figure true
  • Page checklist

    You get

    A draft ready to publish: the query it should win, the title, the links, the call to action, and a source for every figure.

    Why use it

    The page should be written only after you know the site needs it and no existing URL already answers that question.

    Ask when

    • publish a page
    • draft a page
    • build the pages buyers need
    • content checklist
  • One change, then wait

    You get

    One change per URL, the number you had before the change, and a wait before the next edit.

    Why use it

    A title test, a rewrite, and new links on the same URL in the same week make it impossible to know what worked.

    Ask when

    • measure a change
    • set a baseline
    • how long to wait
    • track queries clicks and ranks
  • Weekly and monthly check

    You get

    A weekly or monthly note of what moved in clicks, and which single skill should look at the biggest gap.

    Why use it

    The check reports. It does not edit the site. A real drop goes to the click-drop skill. A quiet week gets one next step.

    Ask when

    • weekly SEO check
    • monthly report
    • repeatable SEO job
    • what should we watch
  • Client memory

    You get

    A per-client log of what changed, what you decided, and what to re-check next time.

    Why use it

    The next session should read the last decision before it recommends something new.

    Ask when

    • what did we change
    • decision log
    • client memory
    • what to recheck

When the situation matches

Open these for a real click drop or a change that moves URLs.

  • Click drop

    You get

    The date the clicks fell, whether it was tracking, the site, or rankings, and a choice to act, wait, or investigate.

    Why use it

    A drop is not a reason to start rewriting. The first result is a diagnosis, not an edit.

    Ask when

    • traffic dropped
    • clicks fell
    • rankings crashed
    • diagnose a drop
  • URL changes

    You get

    A before-and-after URL map and a named way to undo it, confirmed before any redirect goes live.

    Why use it

    Merging, deleting, or moving URLs can throw away rankings if the redirects and the way back are not written down first.

    Ask when

    • redirect
    • merge pages
    • delete a URL
    • domain move
    • CMS migration
    • change URLs

AI recommendations

Citation gaps from data budgetSEO already stores.

  • Gaps in AI recommendations

    You get

    The prompt or keyword, who is cited, the cited URL, and one move: fix the client page or pitch the page that is already cited.

    Why use it

    Clients ask where they show up in AI answers. This uses the citation data budgetSEO already stores.

    Ask when

    • AI Overview gaps
    • who does ChatGPT cite
    • AI recommendations
    • citation gaps
    • Gemini mentions

FAQ

What are budgetSEO SEO skills?

budgetSEO publishes 20 free SEO skills for Claude, ChatGPT, and Cursor. Each skill is a SKILL.md file with the result it produces and the steps to follow. The catalog is https://www.budgetseo.ai/skills. The markdown list is https://www.budgetseo.ai/skills.md.

Are budgetSEO skills free?

The skill pages and SKILL.md files are free to download and install. A lookup that is not already in the account spends the same credits as the app. The skill tells the user the cost and waits for a yes. Reads of saved rankings, quick wins, competitors, and AI citation gaps do not spend lookup credits.

How does a skill reach budgetSEO data?

Skills call the budgetSEO MCP server at https://www.budgetseo.ai/api/mcp. They do not use another SEO data source for keyword, competitor, or domain figures. Connector setup is documented at https://www.budgetseo.ai/mcp.

Where can an agent discover every skill?

The discovery index is https://www.budgetseo.ai/.well-known/agent-skills/index.json. It lists each skill name, description, SKILL.md URL, and SHA-256 digest. https://www.budgetseo.ai/llms.txt links the same files. The AI catalog at https://www.budgetseo.ai/.well-known/ai-catalog.json points at the MCP server card.