SEO Agent Pro

22 agent skills. 614 SEO playbooks. One install command. Works in any AI coding agent.

An SEO skill library for Claude Code, OpenCode, Cursor, Codex, Gemini CLI, Copilot, Amp and anything else that reads markdown — plus knowledge-file packs for ChatGPT and Gemini projects, and .skill bundles for Claude web and Desktop.

It is built on two things most SEO prompt packs do not have:

1. A router that knows what it does not know. It detects your language, your level, your available data and tools, then picks one of five audit modes and labels every claim Confirmed, Likely, Directional, Estimated or Not verified.
2. Execution, not just advice. Data acquisition, audit passes, severity scoring, growth arithmetic, deliverables and client templates — not a list of "10 SEO tips".

Install

# install for every supported agent (default)
npx seo-agent-pro install

# install for one agent
npx seo-agent-pro install --agent opencode
npx seo-agent-pro install --agent claude-code
npx seo-agent-pro install --agent codex
npx seo-agent-pro install --agent agents      # universal .agents/skills alias
npx seo-agent-pro install --agent antigravity
npx seo-agent-pro install --agent gemini-cli
npx seo-agent-pro install --agent cursor
npx seo-agent-pro install --agent copilot
npx seo-agent-pro install --agent amp

# install into the current project only (committed to the repo)
npx seo-agent-pro install --agent claude-code --scope project

# install into any folder you choose
npx seo-agent-pro install --target ./my-seo-skills

# overwrite skills that already exist
npx seo-agent-pro install --agent opencode --force

Then restart your agent and ask an SEO question. It picks up seo-router on its own.

Other commands

npx seo-agent-pro list        # every skill with its one-line description
npx seo-agent-pro info        # package details and file layout
npx seo-agent-pro path        # path to the skills folder
npx seo-agent-pro uninstall --agent opencode
npx seo-agent-pro help

Supported agents

| Agent | Global install | Project install | |---|---|---| | OpenCode | ~/.config/opencode/skills | .opencode/skills | | Claude Code | ~/.claude/skills | .claude/skills | | Cursor | ~/.cursor/skills | .cursor/skills | | Codex | ~/.codex/skills | .codex/skills | | Gemini CLI | ~/.gemini/skills | .gemini/skills | | GitHub Copilot | ~/.copilot/skills | .github/skills | | Amp | ~/.config/amp/skills | .amp/skills | | Any markdown agent | ~/.seo-agent-pro/skills | .seo-agent-pro/skills |

Not a coding agent?

| Platform | How | |---|---| | ChatGPT Projects | Paste chatgpt/00-instructions.md into Instructions, upload the rest of chatgpt/ as knowledge | | Gemini Gems | Upload the 9 files in gems/ | | Claude web / Desktop | Settings → Capabilities → code execution → Skills → upload each claude/*.skill bundle, starting with seo-router |

chatgpt/ and gems/ hold the same library in two different chunkings: ChatGPT Projects takes many small files, Gemini Gems needs fewer, larger ones because of its per-file limit.

What you get

22 skills
└── 614 playbooks, each with the same shape:
    TL;DR → Framework → Actionable checklist → Data points & benchmarks
    → Mistakes to avoid → Key principles

Folder layout

skills/     22 agent skill folders (SKILL.md + articles/), what gets installed
agents/     AGENTS.md + GLOSSARY.md for project-root instruction files
claude/     22 .skill bundles for Claude web and Desktop upload
chatgpt/    15 files for ChatGPT Projects (many small files)
gems/       9 files for Gemini Gems (fewer, larger files)
antigravity/ 9 files for Antigravity surfaces that take documents, not skills
docs/       strategy, audit report, upstream credits, Bangla guides
python/     validator, Traffic Drop Doctor dependencies, operator helpers
site/       landing page and the generated catalogue JSON
tools/      build and verification scripts
_upstream/  unmodified upstream originals, kept for provenance, NOT shipped

Knowledge layer — depth, beginner to expert

| Skill | Covers | |---|---| | seo-router | master entry point: language, level, capability, audit mode, routing, 9 workflows | | seo-fundamentals | how search works, strategy, audits, competitors, updates, penalties, industry playbooks | | keyword-research | discovery, difficulty, volume, intent, mapping, clustering | | on-page-seo | titles, metas, headings, URLs, images, alt text, FAQ, CTR, over-optimisation | | technical-seo | crawling, indexing, robots, sitemap, redirects, canonicals, JavaScript, migrations, international, log files | | link-building | link profiles, outreach, digital PR, guest posts, toxic links, link reclamation | | content-marketing | briefs, calendars, topical maps, decay, pruning, content engineering, distribution | | local-seo | Google Business Profile, Maps, citations, reviews, location pages, multi-location | | enterprise-seo | forecasting, reporting, OKRs, platforms, mergers, at-scale technical | | ai-search | AI Overviews, AI Mode, ChatGPT, Perplexity, Gemini, Copilot, Grok, citations, AI referral tracking | | traffic-drop-doctor | real Python analyser for Search Console exports, plus a client recovery report |

Pro layer — execution

| Skill | Covers | |---|---| | seo-audit-pro | five capability-aware audit modes, page inventory, severity, prioritisation, confidence labelling | | seo-data-ops | Apify-first data acquisition, actor registry, credit-safe runs, native scraping, no-API fallbacks | | seo-keywords-pro | money-keyword selection, keyword-to-page mapping, intent analysis, cluster planning, rank defence | | seo-competitor-pro | competitor discovery, crawling, authority / SERP / content / service-page gap analysis | | seo-backlinks-pro | link quality scoring, link gap, paid-link verdicts, outreach, PR systems, free-platform library | | seo-local-pro | GBP, map-pack audit, NAP citations, review engine, local landing pages, local links | | seo-aeo-geo-pro | AEO/GEO master workflow, direct-answer writing, chunk structure, AI citation optimisation | | seo-entity-reputation-pro | knowledge-graph entity profile, reputation proof stack, third-party corroboration | | seo-strategy-pro | 30/60/90-day roadmaps, query→traffic growth model, content architecture, expectation setting | | seo-verticals | dental, medical, real estate, ecommerce, SaaS, courses, local, agency — niche requirements | | seo-deliverables | client reports, executive summaries, proposals, developer briefs, page briefs, monthly updates, CSV templates |

How it behaves

It tells you the mode before the answer

Audit Mode: Apify Intelligence
Available: URL, target city, competitors, Apify token
Missing: GSC, GA4, GBP, official backlink export
Confidence: strong for crawl, SERP and competitor checks;
            limited for real traffic, conversions and GBP calls.

It never invents data

Missing data is written as a checkable instruction, not a shrug:

Not verified: Backlink count was not available.
Recommended check: Ahrefs → Backlinks → referring domains, or a GSC Links export.

Every recommendation passes a test

Rankings · crawlability · topical authority · local visibility · trust · conversion · AI visibility · risk · measurement · business service. If a recommendation helps none of these, it is cut.

"write high-quality content"
  → "expand /dental-implants/ from 340 to ~1,500 words covering cost,
     candidacy, recovery, risk and financing, add Dentist schema and
     link it from the homepage and the service hub"

It speaks your language

Write in Bangla or Banglish and it replies in Bangla, at your level, keeping SEO terms in English and explaining each one the first time it appears. In non-English markets it researches keywords in the local script, in English, and in transliteration — শাড়ি / saree / sharee — because that is how people actually search.

The Traffic Drop Doctor

The one thing in here that runs code and produces a real artefact.

python skills/traffic-drop-doctor/scripts/analyze_gsc.py \
  --export compare.zip --dates Dates.csv --brand "brand,brand name" \
  --events skills/traffic-drop-doctor/references/google-updates.csv --out report

It splits a click change into an impressions effect and a CTR effect, then labels every losing query and page (ranking loss, CTR loss, demand loss, disappeared). It finds section collapses and migrations by pairing old and new URLs, detects the AI-Overview pattern of CTR loss at stable position, separates brand from non-brand, uses year-over-year data to strip out seasonality, matches the drop date to Google updates, and warns when data is stale or an export is truncated.

Output is a folder: report.md, findings.json, losing-query and losing-page CSVs, section breakdown and a timeline chart. Standard library only; matplotlib is optional.

Validate the library

python python/validate_skills.py skills --strict

Checks frontmatter, links, orphan playbooks, index sync, required sections, raw URLs and leftover vendor names.

Programmatic use

const seo = require('seo-agent-pro')

seo.listSkills()                  // ['ai-search', 'content-marketing', ...]
seo.frontmatter('seo-audit-pro')   // { name, description, version, ... }
seo.listPlaybooks('technical-seo') // ['301-redirects.md', 'accessibility-seo.md', ...]
seo.readPlaybook('technical-seo', 'canonical-tags')
seo.readSkill('seo-router')        // the full router text
seo.glossary()                     // the shared glossary

Credits

This package stands on the shoulders of two projects. Both deserve explicit credit.

Original work — SEO Skills Pack v3.1 by Rashed Hasan Akash

Licensed free to use, adapt and share with attribution; not for resale. Upstream licence preserved at LICENSE.upstream.

Everything below is his original work and is redistributed here unchanged or lightly extended:

• The skill architecture, the router concept, and the multi-skill workflows
• The 11 core skills and their beginner → intermediate → advanced learning paths
• 514 vendor-neutral SEO playbooks, with their consistent structure, checklists, benchmarks and mistake lists
• The Bangla documentation, the Bangla user guide, the offline HTML guide and the prompt library
• Traffic Drop Doctor — the skill, the analyze_gsc.py analyser, the diagnosis guide, the recovery playbooks, the client report template and the demo dataset
• The advanced playbooks (large-site migration, SEO forecasting model, keyword clustering at scale, manual AI visibility audit, title audit at scale, schema markup with JSON-LD)
• The Google update reference data, the validation tooling and the audit process

Original work — Beyond SEO by BeyondTahir / PureDesigners

All Beyond SEO content in this package is credited to BeyondTahir / PureDesigners. Abul Maruf Subono did not author it and does not claim it. It is redistributed here with the author's permission, and was modified and reorganised into the pro skill namespace. The unmodified original is kept for provenance at _upstream/beyond-seo/ and is excluded from the published npm tarball.

This is the data-discipline and execution layer, merged in as the pro skills:

• Capability detection and the five audit modes, with Confirmed / Likely / Directional / Estimated / Not verified labelling
• The no-fluff gate and data quality rules
• Apify-first operation — actor registry, actor shortlists, workflows, credit-safe policy, AI-friendly SEO runs
• Audit execution modules — full site, technical, on-page, content quality, schema, conversion
• Deliverables — client audit report, executive summary, technical, competitor, keyword, backlink and AEO/GEO reports, monthly progress, proposals, developer briefs, page briefs and CSV templates
• Growth planning — 30/60/90-day plans, the 500-query and 5,000-visitor models, content hub and service-page strategy
• Industry verticals — dental, medical, real estate, ecommerce, SaaS/AI, courses, local services, agency
• Entity and reputation SEO — knowledge-graph entity profile, reputation proof stack, third-party authority system
• Backlink execution — quality scoring, link-gap analysis, paid-link verdicts, outreach and digital-PR systems, free platform library
• Entity SEO, reputation SEO, AEO/GEO master workflow, and the local and keyword execution modules

What Abul Maruf Subono built

This is the only part credited to him. He did not write any Beyond SEO content.

• The merge and enhancement of the two projects into one installable library
• A rewritten master router combining language detection, four-level detection, capability detection, audit-mode selection and nine workflows, plus the no-fluff gate and token-safety rules
• 11 new pro skill entry points, each with routing, quick playbooks and guardrails
• The seo-agent-pro CLI, lib/agents.js target registry and index.js programmatic API
• The landing page and the build tooling: catalogue extractor, site builder, upstream audit
• Consolidation of the 10 duplicate GLOSSARY.md copies

Sources

Playbook statistics come from third-party research and public Google documentation, not from original studies. They are written neutrally ("one study found…") without vendor names so they work with any tool. Google update dates come from the Google Search Status Dashboard. Full source notes in docs/CREDITS.md.

Licence

MIT for this repository's own work — the merge and enhancement, the unified 22-skill namespace, the rewritten seo-router, the seo-agent-pro CLI, lib/, index.js, tools/ and the landing page. See LICENSE.

The upstream SEO Skills Pack v3.1 content remains under its author's terms: free to use, adapt and share with credit, not for resale. Upstream text preserved in LICENSE.upstream.

Beyond SEO content is redistributed with the permission of BeyondTahir / PureDesigners. The original is preserved unmodified at _upstream/beyond-seo/.

Nothing in this library guarantees rankings or traffic.

Contributing

git clone <your-fork>
cd seo-agent-pro
npm install          # no runtime dependencies
npm run validate     # python python/validate_skills.py skills --strict

To add a playbook:

1. Create skills/<skill>/articles/<slug>.md using the standard six sections.
2. Link it from that skill's Quick playbooks table and from ARTICLES.md.
3. Run the validator.
4. Open a pull request.

See docs/STRATEGY.md and docs/AUDIT-REPORT.md for the architecture and the upstream audit process.