CiteStat/Wiki/Local scan — checking a site from your own machine

Local scan — checking a site from your own machine

Technical SEO

TL;DR

A local scan fetches your site from your own computer and lets CiteStat judge what it fetched. Use it for a staging site, an intranet page, a localhost dev server, or a site that blocks data-center addresses: our servers cannot or may not reach those, your machine can. The rules, the findings and the score are the same as a normal scan; only the place the requests come from changes.

Explain Simply

A normal scan sends our courier to your shop and looks at what it brings back. A local scan lets you carry the goods to us yourself. We never visit the shop; we only judge what you bring.

One command:

aiseo scan http://localhost:3000 --local
aiseo scan https://staging.example.com --local --ua browser

How it works. The CiteStat command-line program fetches the page from your machine and sends what it got to our server. The server runs the same checks over that capture and answers with what it still needs: robots.txt, the sitemap, a few linked files. The program fetches those too and sends the whole capture again. A typical page needs three or four rounds, and the server stops after five. The last answer is the report.

What goes over the wire. Only what your machine fetched: the status, headers and body of each URL it was asked for. The server never requests your site during a local scan. A file the capture does not hold is read as unreachable.

Who the site sees. Your own address. By default the requests carry our crawler's user agent (CiteStat/1.0 beside AiSEO/2.0), so a robots.txt written for us still applies; --ua browser sends a browser user agent instead.

Private addresses. In a terminal, the host you name may be private (localhost, a staging box). Any other private address is refused, so a page cannot make the program read your network. The MCP tool seo_scan_local refuses private, loopback and link-local addresses on every request, because an AI agent picks its URL from text it read.

What it skips. PageSpeed, domain intelligence, hosting location and blocklist checks ask third parties about your host, so they run only when the host resolves to a public address on our side. The report says when they were skipped.

What it is. The report is labelled "Fetched from your machine" and saved as your own report. It is never a sample report and never a point in the public score history of the domain.

⚡Why It Matters

A scan from our servers cannot see a staging site behind a login or a firewall, a dev server on localhost, or an intranet page, and some sites refuse data-center addresses outright. Those are the places where a missing title, a wrong canonical or a robots.txt that blocks everything is cheapest to fix: before launch. A local scan also lets an AI coding agent check a change and scan again without leaving your machine.

🔍How CiteStat Checks

CiteStat applies its normal rules, tech fingerprints and scoring to the capture the program sent. It does not trust the capture to be complete: what it lacks, the program is asked for in the next round, up to five rounds. A local scan costs one credit when it finishes; the rounds before that are free.

CiteStat Score Impact

The same rules and the same scoring as a server scan. Checks that ask third-party services about your host are skipped for a private or unreachable host, so the number can differ from a server scan of a public copy of the same site. A local scan does not enter the domain's history.

Fix Difficulty

Editor: Easy | Developer: Easy (one command) | AI Agent: Easy (MCP tool or the command line)