What exactly do I get when I buy the SEO agent skill?

You get a folder you download from your library and add to your AI code editor. It contains the workflow your agent follows, two slash commands, all 112 checks with a verification command for each, the scoring method, a report template, and 2 case study teardowns of the products this SEO work was built on.

Which coding agents does the SEO agent skill work with?

The skill works with Claude Code, Cursor, Codex CLI and OpenCode. Claude Code and OpenCode load it as a skill or command folder; Codex reads it through AGENTS.md; Cursor uses project commands or a pasted prompt. Any agent that can read files in a workspace can follow it, because the workflow is plain Markdown rather than a plugin.

Why pay when the playbook and checklist are free?

The playbook explains what to do, and the skill does it. All 20 chapters and all 112 checks with their verification commands are public, so the work can be done by hand. The skill is for when you would rather your coding agent did it: it turns the whole checklist into two commands that audit your site and set up what is missing in your project.

Will the agent change my code without asking?

No. The audit command changes nothing at all: it ends with a report and an ordered plan you approve. The implement command works down that plan one check at a time, verifying each fix and committing it separately with the check id in the message, so any single change can be reverted on its own.

Does the SEO agent skill guarantee better rankings?

SEO for Solos does not guarantee rankings and states plainly that no product can. The skill promises implementation completeness: you will know what is done, what is not, and what each remaining gap costs. Rankings and Core Web Vitals field data are outcomes measured over weeks, and the skill is instructed to say so rather than claim them.

Can I use the skill on client projects?

The licence covers one seat and unlimited projects owned or operated by you or your employer. Client work sits outside it, and an agency licence is not offered. Committing the folder to a public repository is redistribution, which the licence does not permit, so the README explains how to keep it out of one.

Spec 003The product

Your coding agent does the technical SEO

The SEO for Solos agent skill is a folder you plug into Claude Code, Cursor, Codex or OpenCode. It audits your site against 112 technical SEO checks, writes a report with evidence for each one, then sets up everything that is missing in your project, verifying each change as it goes.

Get the skill $49 Audit your site free first

One-time payment · Lifetime updates · 30 day refund, no questions

What happens when you run it

Two commands, in this order. The first one never touches your code.

/seo-audit

Audit and report

The agent maps your existing SEO surface, then works through all 112 checks group by group, running the real verification command for each against your production URL. It records done, todo or not-applicable, with evidence it can quote. A check it could not verify is todo with a note, never done.

It finishes with a scored report and an ordered plan, and stops. Nothing has changed yet.

/seo-implement

Set up and verify

Once you approve the plan, the agent works down it one check at a time. For each one it sets up the fix the way your project is already built, reruns the verification command to prove it worked, and commits with the check id in the message.

One check per commit, so any single change can be reverted on its own when it turns out to be wrong.

What the report looks like

An excerpt from a real audit. The scoring formula is the same one the free checklist uses, so the numbers line up if you also track progress there.

markdown

## Scores

Applicable checks: 104 of 112 (8 marked na: no programmatic tier)
Done: 61   Todo: 43
Percent: 59%   Weighted by impact: 54%

| Group | Done | Todo | na | Percent |
|---|---|---|---|---|
| Foundation | 4 | 1 | 0 | 80% |
| Metadata | 7 | 4 | 0 | 64% |
| Structured data | 6 | 14 | 0 | 30% |
| Crawling and indexing | 11 | 4 | 0 | 73% |
| Programmatic pages | 0 | 0 | 8 | na |
| AI and answer engines | 1 | 6 | 0 | 14% |

## Findings

- found-canonical-host        done   curl -sI https://example.com returns 301 to www
- found-trailing-slash        done   /about and /about/ both resolve to /about, 200
- meta-title-template         todo   every title reads "Example | <page>", brand first
- meta-canonical-self         todo   no canonical on /pricing, /about, /blog/*
- schema-graph-single         todo   3 separate ld+json blocks, no @graph, no @id anchors
- schema-breadcrumb           todo   breadcrumb UI present, no BreadcrumbList markup
- ai-llms-txt                 todo   /llms.txt returns 404
- crawl-robots-ai-split       todo   robots.txt allows all agents, no search/training split
- perf-image-cache-ttl        todo   minimumCacheTTL unset, defaults to 60s

## Implementation plan, by return on effort

| # | Check | Effort | Impact | Planned change |
|---|---|---|---|---|
| 1 | meta-canonical-self | S | 4 | Add self-referencing canonical via metadataBase |
| 2 | meta-title-template | S | 4 | Flip template to "%s | Example", keyword-led default |
| 3 | schema-breadcrumb | S | 4 | Emit BreadcrumbList from the existing breadcrumb data |
| 4 | crawl-robots-ai-split | S | 4 | Generate robots.txt from code, split by bot economics |
| 5 | ai-llms-txt | M | 3 | Build llms.txt from the same config as the sitemap |

## Needs a decision or a person

- crawl-robots-ai-split: which training crawlers to block is your call, not mine.
  The trade-off is citations versus corpus. Chapter: /playbook/bot-economics
- prog-entity-gating: na, this site has no generated page tier.

## Not verified

- perf-lcp-mobile: field data needs weeks of real traffic. Lab number only.

Note the last two sections. A report that marks everything done in one session is a report to distrust, so the skill is instructed to separate what it verified from what needs a human decision and what cannot be checked from a terminal at all.

Install it in your agent

The workflow is plain Markdown, not a plugin. Any agent that can read files in a workspace can follow it; these four have a one-step path.

Claude Code

Skill plus slash commands. Use the home directory path to make it available in every project.

bash

cp -r seo-for-solos ~/.claude/skills/
cp seo-for-solos/commands/*.md ~/.claude/commands/

# then, in any project
/seo-audit

Cursor

Project commands. Keep the folder in the workspace so the agent can read the references.

bash

cp -r seo-for-solos .
mkdir -p .cursor/commands
cp seo-for-solos/commands/*.md .cursor/commands/

# then, in the chat
/seo-audit

Codex CLI

Reads AGENTS.md on every session, so one line is the whole setup.

bash

cp -r seo-for-solos .
echo "For any SEO work, read seo-for-solos/SKILL.md and follow its workflow." >> AGENTS.md

# then ask for an audit in plain words

OpenCode

Same AGENTS.md convention, plus its own command directory.

bash

cp -r seo-for-solos .
mkdir -p .opencode/command
cp seo-for-solos/commands/*.md .opencode/command/

# then, in the session
/seo-audit

What is in the folder

Every count here is generated from the check registry at build time, so this tree cannot claim a file the download does not contain.

text

seo-for-solos/
├── SKILL.md                        the workflow your agent loads
├── commands/
│   ├── seo-audit.md                audit and report, changes nothing
│   └── seo-implement.md            work the plan, one check at a time
├── references/
│   ├── checks.md                   all 112 checks, generated from the registry
│   ├── methodology.md              scoring formulas and sequencing rules
│   ├── report-template.md          the report format to fill in
│   └── case-studies/
│       ├── foreclosuredatahub.md   the large programmatic build
│       └── usagentleads.md         the small-surface build
├── prompts/audit-prompt.md         flat prompt for agents without skills
├── LICENCE.md                      stamped with your email
└── VERSION

112

Checks

Across 12 groups

Coding agents

One-step install each

Commands

Audit, then set up

Case study teardowns

Not published on the site

The part that is not on the site

2 teardowns from products that sold

The public case studies at /results describe what was built. The versions inside the skill go further: they record which decisions actually moved the outcome, which were wasted effort, and what each build got wrong. Your agent reads the relevant one before implementing a group it covers.

Product	Shape	Indexable URLs	Peak MRR
Foreclosure Data Hub	Large programmatic	2,100	$1,500
USAgentLeads	Small surface, full funnel	130	$1,500

Both were built by one person and sold together for $40,000.

What it will not do

It will not guarantee a ranking. Nothing can, and the skill is instructed to say so rather than imply otherwise.
It will not verify Core Web Vitals field data or content quality. Those are outcomes measured over weeks of real traffic, not in a session.
It will not invent a value to fill a gap. A missing description, date or author is reported as a finding, not filled in with something plausible.
It will not rewrite your copy unless you ask. It will flag thin or duplicated content and leave the writing to you.
It will not make the judgement calls for you. Which crawlers to allow and whether a generated page tier earns indexation are decisions the skill surfaces and hands back.

Licence and price

$49, once. One seat, unlimited projects you or your employer own or operate, perpetual, with lifetime updates to this version. Client work sits outside the licence and an agency tier is not offered. The full terms are at /licence.

Get the skill $49 Or score yourself free first

Common questions

What exactly do I get when I buy the SEO agent skill?: You get a folder you download from your library and add to your AI code editor. It contains the workflow your agent follows, two slash commands, all 112 checks with a verification command for each, the scoring method, a report template, and 2 case study teardowns of the products this SEO work was built on.
Which coding agents does the SEO agent skill work with?: The skill works with Claude Code, Cursor, Codex CLI and OpenCode. Claude Code and OpenCode load it as a skill or command folder; Codex reads it through AGENTS.md; Cursor uses project commands or a pasted prompt. Any agent that can read files in a workspace can follow it, because the workflow is plain Markdown rather than a plugin.
Why pay when the playbook and checklist are free?: The playbook explains what to do, and the skill does it. All 20 chapters and all 112 checks with their verification commands are public, so the work can be done by hand. The skill is for when you would rather your coding agent did it: it turns the whole checklist into two commands that audit your site and set up what is missing in your project.
Will the agent change my code without asking?: No. The audit command changes nothing at all: it ends with a report and an ordered plan you approve. The implement command works down that plan one check at a time, verifying each fix and committing it separately with the check id in the message, so any single change can be reverted on its own.
Does the SEO agent skill guarantee better rankings?: SEO for Solos does not guarantee rankings and states plainly that no product can. The skill promises implementation completeness: you will know what is done, what is not, and what each remaining gap costs. Rankings and Core Web Vitals field data are outcomes measured over weeks, and the skill is instructed to say so rather than claim them.
Can I use the skill on client projects?: The licence covers one seat and unlimited projects owned or operated by you or your employer. Client work sits outside it, and an agency licence is not offered. Committing the folder to a public repository is redistribution, which the licence does not permit, so the README explains how to keep it out of one.