What is n-seo?
n-seo is an agentic, local-first control plane for growing organic traffic to your own sites — classic search (SEO), answer engines (AEO), and AI assistants that cite sources (GEO) — without paying an agency to read Search Console for you.
It reads Search Console and GA4 every morning, probes your sites for what quietly broke, and ranks the moves worth making. Your model — Claude, GPT, something local, your choice — turns those findings into proposals and briefings. Your coding agent works the queue over MCP. It proposes; it never posts, publishes, or edits a site.
Why We Built It
Most SEO tools are either a SaaS dashboard that narrates yesterday’s charts, or an agency retainer that does the same reading for a monthly fee. Neither leaves you with a ranked queue you own, on hardware you control, with rules you can read and change.
We built n-seo as the generalized version of the control plane En Dash already runs on its own portfolio every morning. Pulls and probes are plain deterministic code. Judgment goes to a model you configure. Anything that changes a site stays with you — in your repo, on a branch, the way you already work.
A Morning Loop, Not Another Dashboard
Four planes. The first two are code; the third is you and your agent; the fourth is what keeps the second one honest.
- Awareness — Search Console (16 months and the trailing 90 days), GA4, URL Inspection verdicts, and a no-auth probe of each site: robots, sitemap, soft 404s, blocked AI crawlers, JS-only shells.
- Analysis — Rules turn snapshots into actions: metadata that misses the ranking language, pages that rank but rarely get clicked, queries at position 5–15, landing pages that don’t deliver. Impact per unit of effort orders the queue.
- Action — Each card opens to the evidence, the move, and a build spec. You ship. Machine proposals wait in a holding area until you accept them.
- Measurement — Shipped work becomes watching, never deleted. The next morning’s data says whether it worked.
npm install -g n-seo, then n-seo init my-sites. Five minutes on demo data, no Google setup required.
Built Beside Your Sites, Not Inside Them
n-seo sits next to your repos. It reads your sites the way Google sees them — through Search Console and GA4 — so it never needs a copy of their code. n-seo init makes one directory of its own; everything it writes stays there. It refuses to initialize inside a site repo.
- Your model, whichever you pick. Anthropic, any OpenAI-compatible endpoint, or a CLI on your machine (claude -p, ollama run). No key ships with n-seo. Turn the LLM module off and you still get the full data-derived queue.
- Your coding agent, on the queue. A read-only MCP server exposes the queue, metrics, index coverage, and ops health. Seven skills (orient, setup, add-site, triage, ship, review, deploy) install into the instance so setup and shipping are things you ask for in plain language — with freezes and weekly metadata budgets enforced, not merely documented.
- Yours, on your hardware. Laptop, VPS, NAS, or a VM in your own cloud. No account, no SaaS, no telemetry. MIT licensed.
Useful for People Who Own the Sites
This is for founders, indie hackers, and small teams who ship their own sites and are tired of either guessing or outsourcing the reading of Search Console.
If you want someone else to rewrite titles and post community comments for you, this is the wrong tool on purpose. If you want a ranked queue with evidence, a model that drafts proposals, an agent that can ship under your rules, and a morning log that tells you whether last week’s change moved anything — start here.
Built to Make the Idea Click
Some SEO advice fits in a checklist. Running a portfolio needs a loop that runs every morning and remembers what you already tried.
n-seo exists so more people can see the next move with the numbers behind it, hone their own principles instead of renting someone else’s, and keep the words — and the deploys — always theirs.
Free means free. Same shop as n-dx, Hotdog Benchmark, and the rest of the En Dash Toolkit.