# Inspector
Inspector is a one-shot, on-demand audit for any URL on the public internet. Paste a URL, pick mobile or desktop, and Inspector fans out to six checks, computes a single 0–100 **Inspector Score**, and lists prioritized issues with concrete recommendations.
Unlike RUM or Synthetic Monitoring, Inspector does **not** require the URL to be a monitored domain in your workspace. It's a quick second opinion on any page - your own, a competitor's, a staging environment, or a page you're about to ship.
## Where to Find It [#where-to-find-it]
Inspector is a top-level tab in the header. Clicking it opens the Inspector workspace at `/dashboard/inspector`.
## Running an Inspection [#running-an-inspection]
The input page has a single hero card with:
* A URL input - placeholder `https://www.example.com/page`. Must be `http://` or `https://`, public, and resolvable. Localhost, private IPs, and link-local addresses are rejected.
* A **device toggle** - two icon buttons next to the URL: phone (mobile) and monitor (desktop). The headless Chrome check runs in the matching profile, and CrUX field data is requested for the matching form factor.
* A **Run inspection** button - disabled while a run is in flight or your quota is exhausted.
After you submit, the page redirects to the detail view (`/dashboard/inspector/[testId]`) where progress and results render in place. Every test gets a stable URL from the moment it starts.
### Quota [#quota]
Each workspace has a monthly Inspector quota set by its plan:
| Plan | Inspector tests / month |
| -------- | ----------------------- |
| Free | 3 |
| Starter | 10 |
| Lite | 30 |
| Standard | 100 |
| Advanced | 500 |
The quota chip in the top-right corner of the input page shows current usage:
* Available: `5 / 30 inspections · Resets Apr 20`
* Exhausted: `Quota exhausted · Resets Apr 20` - the **Run inspection** button is disabled until reset.
The cycle is anchored to your workspace's package activation day, not the first of the calendar month. A workspace activated on the 14th resets on the 14th of every month. Current usage is also shown on the workspace [Billing & Plans](/workspace/billing) page alongside your other resource limits.
A test counts as one quota unit regardless of which checks succeeded. Tests that fail URL validation (invalid scheme, private IP, unresolvable host) do not count.
## History Sidebar [#history-sidebar]
The left-hand sidebar lists past inspections for the workspace. Each row shows:
* A device icon (mobile / desktop)
* The inspected URL
* A status badge (queued / running / completed / failed)
* A timestamp
A **Filter by URL…** input narrows the list. Empty states read **No inspections yet.** or **No matches.**
## What Happens After You Submit [#what-happens-after-you-submit]
Inspector runs six checks in parallel:
| Check | What it captures |
| -------------- | ----------------------------------------------------------------------------------------------------- |
| **CrUX** | Real-user p75 Core Web Vitals from Chrome users (LCP, INP, CLS, FCP, TTFB) |
| **Chrome** | Headless-Chrome lab audit - full waterfall, filmstrip, console errors, SEO snapshot, security headers |
| **Uptime** | HTTP status, redirect chain, TTFB, total response time, response headers |
| **SSL** | Certificate validity, expiry, chain completeness, key strength, TLS version |
| **WHOIS** | Registrar, creation date, expiry, nameservers, privacy flag |
| **robots.txt** | Raw file, parsed groups, and a clear allow/disallow verdict for the inspected URL |
Each check runs independently and is retried once on transport failure. A check can finish in three states: **success**, **no\_data** (e.g. CrUX has no data for low-traffic URLs - not an error), or **failed**.
While the test runs, the detail page shows a progress bar with `~Xs remaining` based on the checks not yet finished. When all checks have settled, the label briefly switches to **Finalizing…** while the Inspector Score and issues are computed.
If you submit a test and the queue isn't empty, you'll see `Queued - N tests ahead of you` until your run starts.
## What You See When It's Done [#what-you-see-when-its-done]
The detail page renders, top to bottom:
* A **Verdict banner** - the device badge (purple for mobile, blue for desktop) and the inspected URL, followed by a plain-English verdict headline that a non-technical owner can act on. When AI phrasing is enabled it also carries a short 1–2 paragraph narrative marked **Generated by AI**, plus a row of stat pills (page weight, image share, the worst single script's blocking time, and request count).
* The **Inspector Score** card - the 0–100 score in a gauge, the count of critical / warning / info issues, and a per-pillar breakdown (Performance, SEO, Security, Infrastructure, Availability) showing each pillar's percentage and points.
* Per-check sections rendering each successful check's data - Overview, Field data (CrUX), Lab data (Chrome vitals), Visual Loading, Waterfall, CWV Diagnostics, Resources, SEO, Crawlability, and Security.
* The **Issues** panel - every finding the rule engine produced, each with a severity, category, description, recommendation, and a link to the relevant report section.
* A **section navigation pill** pinned to the bottom of the viewport, jumping you between Overview, Field data, Lab data, Visual Loading, Waterfall, CWV Diagnostics, Resources, SEO, Crawlability, Security, and Issues.
If a previous inspection of the same URL and device exists, key metrics carry a small **delta badge** showing the change vs the prior run.
Inspector results are kept for 12 months. Past tests remain visible in the history sidebar until they expire.
## What Inspector Does Not Do (Yet) [#what-inspector-does-not-do-yet]
* No scheduled re-runs - Inspector is one-shot. Use Synthetic Monitoring for recurring tests.
* No public share links or PDF export.
* No diff view between two runs (delta badges aside).
* No retry button on a failed check - re-submit the test.
## Related [#related]
How the 0–100 score and grade are computed.
What each of the six checks measures.
Severities, categories, and how findings are deduplicated.
Recurring Lighthouse tests for monitored URLs.