SEO & AI Search

Site Audit

Most sites lose rankings to unglamorous problems: a slow largest-contentful-paint, missing meta descriptions, two H1s on one page, no sitemap. Individually trivial, collectively decisive — and invisible unless something checks for them. The Site Audit checks for them.

Running an audit

Open SEO → SEO Audit, enter a URL, and Start audit. Audits run in the background (statuses: queued, running, complete) and finish with a notification. Once a week the system audits your site by itself, so drift gets caught without ritual.

Reading the report

Audit Score — the single number to watch over time.
Metric tiles — mobile performance, LCP, total blocking time, and total issue count.
AI summary — a plain-language read of what matters most in this run.
Issues — each finding with severity and a How to fix drawer written for humans, not just for SEOs.
Passed checks — what's already right, so you know what not to touch.
Per-page metadata table — one row per page: word count, H1, title, meta description, Open Graph, canonical, schema, image alt coverage, and lang attribute. This table is where thin or half-configured pages hide.
Audit history — every run with score and status, so improvements (and regressions) are visible as a trend.

What gets checked

Around two dozen checks in four families:

Performance — mobile score, render-blocking resources, LCP, CLS, total blocking time.
On-page — unique document titles, meta descriptions, H1 presence, heading order, image alt text, html lang, canonical self-reference, viewport, link names.
Structure & crawlability — sitemap.xml, robots.txt, llms.txt, crawlable anchors, internal linking, content depth.
Trust — HTTPS and browser best practices.

Note that llms.txt appears here as one check among many; the dedicated AI Visibility audit goes much deeper on the AI side.

Fix all with AI

The report's distinguishing feature: findings are not just described but actionable by the agent. Fix all with AI hands every issue to the agent in one session, ordered by severity, and the agent edits your actual code — adding the missing meta descriptions, fixing the heading order, generating the sitemap.

Requirements and behavior:

Works in the desktop app with a bound project folder (you'll be prompted to bind one if you haven't).
Every change lands as a reviewable diff in chat. Nothing is committed behind your back; you keep or revert each edit.
You can also fix issues one at a time from each finding's drawer, or ask in Chat for any subset ("fix just the image alt issues").

Anonymous audits

The audit engine also powers the free scanner on the goshipfast.com homepage — paste any URL, no account needed. The in-product version adds history, scheduling, the metadata table, and the fixing agent.

Next instrument: AI Visibility, the same rigor pointed at AI search.