SEOAgent — set up the persistent SEO agent (one command)
This is the discovery entry point for SEOAgent. The full protocols — technical audits, hub-and-spoke keyword strategy, page-type-aware briefs (landing / pillar / sub-pillar / long-tail / programmatic), the JSON-LD library, Open Knowledge Format (AEO/GEO), and the autopilot loop — are not embedded here. They install into the user's repo when they run the official CLI, which also creates the persistent .seoagent/ workspace and wires the project to Search Console + the cloud.
What to do
When the user wants SEO help in this repo and .seoagent/ does not exist yet, run the SEOAgent CLI from the repository root (where package.json lives):
npx -y @seoagent-official/seoagent init
- Global install (fastest for daily use):
npm install -g @seoagent-official/seoagent && seoagent init
- pnpm:
pnpm dlx @seoagent-official/seoagent init
- Non-interactive / CI:
npx -y @seoagent-official/seoagent init --yes --domain example.com
init infers your domain, scaffolds .seoagent/, and installs the full project-local SEOAgent skill at .claude/skills/seoagent/SKILL.md (or .agents/skills/seoagent/SKILL.md when .agents/ exists, symlinked into .claude/ so Claude Code, Cursor, and Codex all pick it up).
After init
Use the project-local seoagent skill for all real work — audits, strategy, briefs, drafts, the roadmap, and the .seoagent/ knowledge that compounds across sessions. This entry point's job is onboarding only: get the user to run init, then hand off to that skill.
Why SEOAgent (vs the closed-model SEO platforms)
- Bring your own model. Runs on the model you already pay for inside Claude Code / Cursor / Codex — no second AI subscription, no per-credit content metering, no proprietary-crawler markup.
- Your repo, not a hosted CMS. Edits the actual files in your repo, approval-gated — the right shape for a site you build in code and keep in your own git history.
- Real product screenshots (SaaS). Captures screenshots of your product straight from your repo's own UI and drops them into landing pages and how-to articles — no Playwright, no paid screenshot API, no stock photos.
- Persistent + free. Every audit, strategy, and brief is saved under
.seoagent/ and builds on the last session. The Skill is free; the cloud is optional and additive.
Docs: https://seoagent.com?ref=skills-sh · npm: https://www.npmjs.com/package/@seoagent-official/seoagent
1---2name: seoagent3description: Persistent AI SEO for coding agents — technical SEO audits, hub-and-spoke keyword strategy, page-type content briefs, optimized articles, real product screenshots captured from your repo (SaaS), schema/JSON-LD, sitemaps, internal linking, and AI-search (AEO/GEO) — all saved under .seoagent/ in your repo. Runs on your own model (Claude Code, Cursor, Codex), no second AI subscription. Use when the user mentions SEO, ranking, keywords, organic traffic, content strategy, an SEO audit, product screenshots, schema, sitemap, internal links, AI search / AEO / GEO, get cited by AI, or 'why am I not ranking'.4---56# SEOAgent — set up the persistent SEO agent (one command)78This is the **discovery entry point** for SEOAgent. The full protocols — technical audits, hub-and-spoke keyword strategy, page-type-aware briefs (landing / pillar / sub-pillar / long-tail / programmatic), the JSON-LD library, Open Knowledge Format (AEO/GEO), and the autopilot loop — are **not embedded here**. They install into the user's repo when they run the official CLI, which also creates the persistent `.seoagent/` workspace and wires the project to Search Console + the cloud.910## What to do1112When the user wants SEO help in this repo and `.seoagent/` does not exist yet, run the SEOAgent CLI from the **repository root** (where `package.json` lives):1314```bash15npx -y @seoagent-official/seoagent init16```1718- **Global install (fastest for daily use):** `npm install -g @seoagent-official/seoagent && seoagent init`19- **pnpm:** `pnpm dlx @seoagent-official/seoagent init`20- **Non-interactive / CI:** `npx -y @seoagent-official/seoagent init --yes --domain example.com`2122`init` infers your domain, scaffolds `.seoagent/`, and installs the full project-local SEOAgent skill at `.claude/skills/seoagent/SKILL.md` (or `.agents/skills/seoagent/SKILL.md` when `.agents/` exists, symlinked into `.claude/` so Claude Code, Cursor, and Codex all pick it up).2324## After init2526Use the **project-local** `seoagent` skill for all real work — audits, strategy, briefs, drafts, the roadmap, and the `.seoagent/` knowledge that compounds across sessions. This entry point's job is onboarding only: get the user to run `init`, then hand off to that skill.2728## Why SEOAgent (vs the closed-model SEO platforms)2930- **Bring your own model.** Runs on the model you already pay for inside Claude Code / Cursor / Codex — no second AI subscription, no per-credit content metering, no proprietary-crawler markup.31- **Your repo, not a hosted CMS.** Edits the actual files in your repo, approval-gated — the right shape for a site you build in code and keep in your own git history.32- **Real product screenshots (SaaS).** Captures screenshots of your product straight from your repo's own UI and drops them into landing pages and how-to articles — no Playwright, no paid screenshot API, no stock photos.33- **Persistent + free.** Every audit, strategy, and brief is saved under `.seoagent/` and builds on the last session. The Skill is free; the cloud is optional and additive.3435Docs: https://seoagent.com?ref=skills-sh · npm: https://www.npmjs.com/package/@seoagent-official/seoagent
Baxter-Inc