one-step-seo · git:20261108.80f295b · 2026-11-08 · sha256 8225ba8a0f2e2bb0
one-step-seo git:20261108.80f295bA
Immutable. This exact content never changes and is served at /api/v1/blob/8225ba8a0f2e2bb0.
---
name: one-step-seo
description: One-command SEO + AI-visibility audit for any website. Use when asked to audit SEO, check rankings readiness, validate schema, improve AI citation, or plan SEO fixes.
---
# one-step-seo skill
Run a full SEO + GEO/AEO audit in one step. Zero dependencies, Node 18+.
## When to use
- User asks: "audit my SEO", "why aren't we ranking", "check this page", "fix schema", "AI search readiness", "GEO audit".
- Example prompt to support: "Audit https://example.com with one-step-seo, summarize the top 5 fixes, and create an implementation plan."
- Any framework: Astro, Next.js, WordPress, Shopify, static HTML.
## Workflow
1. Run the audit (never guess — always fetch the live URL):
```bash
# Auto-discover all website pages and generate full report:
npx one-step-seo https://example.com
# or dedicated site command (hybrid sitemap + link BFS crawl):
npx one-step-seo site https://example.com
# or full crawl with custom page limit/out folder:
npx one-step-seo audit https://example.com --all --out ./seo-report
```
Beginner shortcut (5 pages, all formats):
```bash
npx one-step-seo quick https://example.com
```
Single page:
```bash
npx one-step-seo page https://example.com/pricing --out ./seo-report
```
CI gating (exit 2 on P0 or worse):
```bash
npx one-step-seo audit https://example.com --all --out ./seo-report --fail-on P0
```
Offline local builds (Astro, Next.js static, Vite, HTML):
```bash
npx one-step-seo site ./dist --out ./seo-report
npx one-step-seo page ./dist/index.html --out ./seo-report
```
Inspect auditing skills & filter by skill domain:
```bash
npx one-step-seo skills
npx one-step-seo audit https://example.com --skills onpage-optimization,schema-validation
```
2. Read `./seo-report/report.json` (machine) or `report.md` (human). Multi-page runs also write `report-N.*`, `index.md`, `site-report.json`, and a site-wide `index.html` dashboard (totals, worst-first pages, recurring issues, category distribution, duplicate titles/metas, internal linking & orphan risks) — start there for the big picture.
3. Present the **two scores separately** — Search SEO 0-100 and AI Visibility 0-100, never blended.
4. Fix in priority order P0 → P1 → P2. Each fix needs evidence from the report.
## Specialized Agent Sub-Skills (`skills/*/SKILL.md`) vs Executable CLI Skill Groups
`one-step-seo` separates **24 Agent Skill playbooks** (`skills/*/SKILL.md` across 10 domains — structured instructions for AI coding agents) from **22 executable CLI skill groups** (`lib/skills.mjs` across 9 categories — rule filters passed to `--skills` / `--exclude-skills` and inspected via `one-step-seo skills`).
Delegate domain-specific engineering and editorial workflows to the 24 structured playbooks in `skills/`:
| Domain | Sub-Skill Path | Primary CLI Commands & Flags | When to Activate |
| -------------------------- | -------------------------------------------- | -------------------------------------------------------------- | --------------------------------------------------------------------------------- |
| **Core Workflow** | `skills/seo-audit/SKILL.md` | `audit`, `page`, `quick`, `ai` | Full-site or single-page audits, score triage, and 30/60/90-day action plans |
| **Core Workflow** | `skills/seo-fix/SKILL.md` | `fix [--apply] [--url] [--lang]` | Safe, dry-run-first automated HTML remediation with `.bak` backups |
| **Technical SEO** | `skills/technical-seo/SKILL.md` | `audit/page --skills technical-seo` | HTTPS, HTTP 200 status, redirect chains, `noindex`, canonical URLs, mixed assets |
| **Technical SEO** | `skills/http-status/SKILL.md` | `audit/page --skills http-status [--check-links]` | 4xx/5xx responses, multi-hop redirects (`>1` hop), MIME headers, broken outlinks |
| **Technical SEO** | `skills/javascript-seo/SKILL.md` | `audit/page --skills javascript-seo` | CSR vs SSR/SSG raw-HTML blindness, empty SPA shells, script & hydration bloat |
| **Technical SEO** | `skills/sitemap-robots/SKILL.md` | `sitemap`, `audit --skills sitemap-robots --crawl sitemap` | `robots.txt` directives, XML sitemaps, `<sitemapindex>` expansion |
| **Security & Hygiene** | `skills/security-hygiene/SKILL.md` | `audit/page --skills security-hygiene --block-private-ips` | HTTPS transport, active/passive mixed content, CSP/HSTS headers, SSRF safety |
| **Accessibility & UX** | `skills/accessibility-seo/SKILL.md` | `audit/page --skills accessibility-seo`, `fix --only lang` | `<html lang>`, mobile viewport, image `alt`, heading order, `<main>`, anchors |
| **International & Local** | `skills/international-seo/SKILL.md` | `audit/page --skills international-seo` | `hreflang` clusters, `x-default`, BCP-47 `lang`, self-referencing locale URLs |
| **International & Local** | `skills/local-seo/SKILL.md` | `page --skills local-seo`, `schema --generate localbusiness` | `LocalBusiness` JSON-LD, Name-Address-Phone (NAP) consistency, geo/hours |
| **On-Page & Content** | `skills/onpage-optimization/SKILL.md` | `audit/page --skills onpage-optimization` | `<title>`, meta description, H1–H6 hierarchy, Open Graph, Twitter Card, alt/CLS |
| **On-Page & Content** | `skills/social-metadata/SKILL.md` | `audit/page --skills social-metadata`, `fix --only og,favicon` | Rich Slack/LinkedIn/X link previews (`og:*`, `twitter:card`, `og:url`, favicons) |
| **On-Page & Content** | `skills/image-seo/SKILL.md` | `audit/page --skills image-seo` | Image `alt` descriptions, `width`/`height` CLS prevention, image count & HTTPS |
| **On-Page & Content** | `skills/content-quality/SKILL.md` | `audit/page --skills content-quality [--profile ...]` | Word count depth, answer-first lead blocks, comparison tables (`<th>`), FAQs |
| **On-Page & Content** | `skills/duplicate-content/SKILL.md` | `audit --pages N --skills duplicate-content` | Cross-page duplicate titles/descriptions and canonical consolidation |
| **On-Page & Content** | `skills/internal-linking/SKILL.md` | `audit --pages N --skills internal-linking [--check-links]` | Crawl depth, orphan/dead-end pages, generic anchor text, broken same-host links |
| **GEO / AEO / AI Search** | `skills/geo-optimization/SKILL.md` | `ai`, `audit --skills geo-optimization` | AI crawler access (`GPTBot`, `ClaudeBot`, `PerplexityBot`), snippet gates, facts |
| **GEO / AEO / AI Search** | `skills/answer-engine-optimization/SKILL.md` | `page/ai --skills answer-engine-optimization` | 40–60w answer capsules, question H2s, `<th>` tables, quantified stats, FAQ parity |
| **GEO / AEO / AI Search** | `skills/entity-optimization/SKILL.md` | `page/ai --skills entity-optimization`, `schema` | Knowledge Graph entity alignment (`<title>`/`<h1>`/`og:title`), authorship, dates |
| **GEO / AEO / AI Search** | `skills/llms-txt/SKILL.md` | `llms`, `audit --skills llms-txt` | `/llms.txt` discovery, structural quality (`G05`), starter draft generation |
| **Structured Data** | `skills/schema-validation/SKILL.md` | `schema [--generate <kind>]`, `audit --skills schema` | JSON-LD syntax, `@graph` entity linking, 15 required-property validators, parity |
| **Structured Data** | `skills/ecommerce-seo/SKILL.md` | `audit/page --profile product --skills ecommerce-seo` | `Product`/`Offer`/`Review`/`BreadcrumbList` JSON-LD, facet canonicals, PDP media |
| **Performance & Quality** | `skills/core-web-vitals/SKILL.md` | `audit/page --psi [--psi-strategy desktop] --skills cwv` | HTML payload size, image count, Lighthouse lab metrics (LCP/CLS/TBT) & CrUX INP |
| **Monitoring & Reporting** | `skills/seo-regression/SKILL.md` | `diff <base.json> <head.json> [--fail-on P0\|P1]` | CI/CD regression gating, score deltas, `rulesVersion` verification |
## Reading report.json
Stable, schema-documented shape (`lib/schema-report.json`):
- `scores.search {score, band}` and `scores.ai {score, band}` — the two axes. Bands A–F.
- `counts {P0, P1, P2, P3, pass}` — severity totals for gating and progress tracking.
- `findings[] {id, category, severity, title, evidence, fix, seoImpact, geoImpact}` — priority-sorted, P0 first. Finding IDs match `^[A-Z]+[0-9]+-[a-z0-9-]+$`.
- `page {title, metaDescription, canonical, wordCount, h1[], images, internalLinks, externalLinks, psi?}` — page facts without re-fetching.
- `site {origin, robots, sitemap, sitemapUrls, llms}` — booleans + sitemap size.
- Severity: P0 critical (today) → P1 high (this week) → P2 medium (this month) → P3 polish → pass healthy.
- Categories: `technical`, `onpage`, `content`, `schema`, `geo`, `performance`.
## Prioritizing fixes
1. Take the first 3–5 non-`pass` findings (they are already priority-sorted).
2. For each: quote `evidence`, state `fix`, and add triage from `report.md` (“Fix this first”): impact (low/medium/high), effort (minutes/hours/days), owner (developer/content/SEO/agency).
3. Group by owner so the user can delegate: content fixes to writers, developer fixes to engineers.
4. Never present a fix without its evidence. Never invent statistics, reviews, business facts, or competitor data.
## Remediation plans
Structure every plan as:
- **Executive summary** (3 bullets: scores, verdict, biggest win).
- **Priority actions** (Problem → Evidence → Action → Impact → Effort → Owner).
- **Passed checks** (1 line each — reassure the user what is fine).
- **30/60/90-day plan** (P0/P1 now, P2 this month, P3 + monitoring later).
- **Verification**: re-run `npx one-step-seo page <url>` after fixes and report the score delta.
## Rules
- Never invent statistics, reviews, business facts, or competitor data.
- Only recommend Schema.org markup for content visible on the page.
- Prefer additive, reversible fixes. Show the exact snippet or diff.
- Distinguish: **verified fact** vs **observed data** vs **inference** vs **recommendation**.
- Never claim guaranteed rankings or guaranteed AI citations. Frame results as readiness signals and best practices.
## Explaining findings to non-technical users
- Lead with the verdict sentence from `report.md` (“Executive summary”).
- Translate each finding via its plain-language explanation (same section): no jargon, one sentence on why it matters to traffic or customers.
- Use impact/effort/owner labels so they know what to approve, how long it takes, and who does it.
- End with the citable checklist (“Make this page more citable”) for AI-visibility work.
## Client-ready summary
```markdown
## SEO audit — <url>
- Search SEO: **{search}/100 ({band})** · AI Visibility: **{ai}/100 ({band})**
- Verdict: {one-sentence verdict from report.md}
- Fix this first:
1. [{sev}] {title} ({id}) — {one-line fix} (Impact: {x} · Effort: {y} · Owner: {z})
2. …
- Passed: {count} checks, including {2–3 notable passes}.
- Next: {re-run command} after fixes. Full report: `report.html`.
```
## Report shape
`report.json` fields: `scores {search {score,band}, ai {score,band}}`, `rulesVersion` (scoring ruleset — diff warns across versions), `counts`, `findings[] {id, category, severity, title, evidence, fix}`, `page`, `site`.
Severity: P0 critical, P1 high, P2 medium, P3 low, pass healthy.
## Fix workflow
For safe auto-fixes on **local HTML files**, prefer the built-in writer:
```bash
npx one-step-seo fix ./dist --url https://example.com/ # dry-run diff
npx one-step-seo fix ./dist --url https://example.com/ --apply # write + .bak
```
It is additive-only, previews a unified diff, verifies every fix in memory,
and never invents facts (language/titles/images need flags when they cannot
be derived from the page). For everything else, follow
`skills/seo-fix/SKILL.md`: default to preview and ask before writing files.
## AI-visibility paperwork
```bash
npx one-step-seo llms https://example.com # llms.txt status + starter draft + robots snippet
```
Review the starter draft with the user before publishing — it contains
explicit placeholders where facts are unknown. The robots snippet allows the
major AI crawlers; only apply it when AI citation is wanted.