Changelog

Something shipped, and your site got slower. Which release was it, and what broke?

Deploys shows where it went wrong. Making it easier to find and fix regressions, and track new releases.

Now in Real User Monitoring and Synthetic, it compares every release with the one before it, flags the ones that regressed, and shows which metrics and pages changed.

Select a release to see which metrics got worse, by how much, and which pages were most affected. Improvements are listed too.

Pick a metric (UX Rating by default in RUM, Performance Score in Synthetic), or filter to Regressed to see only the releases that got worse.

• RUM shows how each release changed the experience for your real visitors.
• Synthetic uses your scheduled tests, with a line per Test Profile, so you can see whether mobile, desktop or both got worse. Every test runs on the same device and connection, so any change points to the release itself.

Find it at Site → RUM → Deploys or Site → Synthetic → Deploys. Read more in the RUM Deploys and Synthetic Deploys docs.

Ask your coding agent#

Calibre MCP can check deploys too, using RUM or Synthetic data. Ask “Did yesterday’s deploy make anything slower?” and your agent compares it with the deploy before.

Getting started#

Deploys are included on every plan. Send deploys from your CI, or connect GitHub and Calibre records them for you.

Guide links have had a small design refresh. They now sit in the same spot on every page that has one: the top right, next to the page title.

So whether you’re on the RUM Dashboard, Third Parties, CrUX Pages or your Test Scheduler, you’ll know where to find help for the page you’re on. Each link opens the matching docs page in a new tab.

Calibre MCP connects your coding agent to your RUM, CrUX and Synthetic data. Ask about any page in plain language. Your agent finds the slow element, script or deploy, and can then fix it in your codebase.

• “Why is LCP slow on /checkout?”
• “Fix the slowest interaction on /search.”
• “Did yesterday’s deploy make anything slower?”

It works with Claude Code, Cursor, Codex and GitHub Copilot, and with Claude and ChatGPT. There’s nothing to install, and it’s included on every plan.

claude mcp add --transport http calibre https://calibreapp.com/mcp

// .cursor/mcp.json
{
	"mcpServers": {
		"calibre": {
			"url": "https://calibreapp.com/mcp"
		}
	}
}

codex mcp add calibre --url https://calibreapp.com/mcp

// .vscode/mcp.json
{
	"servers": {
		"calibre": {
			"type": "http",
			"url": "https://calibreapp.com/mcp"
		}
	}
}

Other agents connect to https://calibreapp.com/mcp.

Your agent only sees what you can see#

• Read-only: your agent can read the Sites your Calibre account can see, and can’t change anything.
• Admins stay in control: turn MCP off for your organisation in Manage → Settings, and review or revoke connections in Manage → Connected apps.
• Connections are recorded: on the Company plan, the audit log records when apps connect or disconnect, and changes to AI assistant access.

We don’t and won’t train AI models on your data, and we don’t send it to AI providers. MCP is an API, here if you want it: your data only reaches an AI tool when you connect one.

We’ve updated our Privacy Policy, Terms of Service, Data protection and Security pages to cover connected apps.

Read the MCP documentation to set up your agent.

Our Test Agents are now using Chrome 156 and Lighthouse 13.6.0 to test your Sites.

Chrome 156 fixes CVE-2026-118320, a high severity memory corruption vulnerability in V8.

LCP, FCP and Performance Score changes#

Lighthouse 13.6.0 fixes how simulated throttling calculates Largest Contentful Paint and First Contentful Paint:

• Text LCP: when the LCP element is text painted at the same time as FCP, LCP now equals FCP. Previously, preloaded and other non-blocking requests could push LCP later. This was reported in June 2025.
• Render-blocking requests: Lighthouse now uses Chrome's own render-blocking data instead of guessing from request priority, so requests that don't block rendering, including module scripts, no longer count against FCP and LCP (#16869, #17163).

Expect LCP and FCP to mostly improve on PageSpeed Desktop and PageSpeed Mobile Test Profiles, lifting Performance Scores. Test Profiles using runtime throttling aren't affected by these fixes.

New: Agentic Browsing audits#

Snapshots now include Lighthouse's Agentic Browsing audits, which check how well AI agents can navigate and use your pages:

• Accessibility tree: whether it's well-formed.
• WebMCP: registered tools, form coverage and schema validity.
• llms.txt: whether your file follows community recommendations.
• ARD manifest: whether your Agentic Resource Discovery manifest is valid.

Google marks this category as under development, so expect the audits to change. Read more in the Lighthouse Audits documentation.

Today, Google announced four experimental ad metrics in the Chrome User Experience Report (CrUX). They’re now in our free Core Web Vitals Test, in a new Advertising section.

Here’s a live report for independent.co.uk to dig into while you read.

The four metrics are:

• Viewport ad density: the share of the viewport covered by ads.
• Viewport ad count: the number of ads visible in the viewport.
• Total CPU usage by ads: processor time used by ad frames.
• Total network usage by ads: data downloaded by ad frames and ad resources.

To be eligible for ad metrics data in CrUX, a site needs an ads.txt file with at least one authorised seller. If there’s no ad metrics data for a site, the Advertising section simply isn’t shown.

Read the ad metrics documentation for the full detail.

Today we are bringing Deployments to Real User Monitoring (RUM), making it easier compare performance before and after a release.

In RUM, you’ll see deployments running across the bottom of charts. Deployment flags display on hover and include commit details. You can click through to see the commit in your source control system.

Each metric page now includes a segmented view, so you can see how recent deploys compare, with full context of how much data is being assessed.

Viewing more detail allows you to see recent deployments over time, and compare longer-term changes.

Getting started with Deployments#

Deployments are available on all plans, and any deploys you’ve already sent to Calibre will appear on your charts already.

You can notify Calibre when you deploy your website using webhooks, the Calibre CLI, or the Node.js API.

Or, alternatively, connect your GitHub account from Site → Settings → Integrations, and Calibre will watch GitHub for production deploys, using the Deployment Branches setting to work out which deployments count as production.

Synthetic Pulse and Metrics have been revamped to display median aggregated metrics, combined for all pages.

Previously, you would view Synthetic metrics for a single page at a time. Now you can view all pages at once, or filter by a single page, as before.

Each metric page includes a segment views for:

• By Test Profile: how each device profile differs.
• By Deployment: how deployment changed the metric.
• By Page: whether a change is site-wide or isolated.

Navigate to Synthetic → Pulse and click on a metric title to see the new metric pages, or go direct to Synthetic → LCP from the left-hand navigation to see the new metric page for Largest Contentful Paint.

You can freely switch between metrics, pages, and date ranges to see how your site performance is trending over time, and see how each segment contributes to performance.

Monitoring in-page UX in single-page apps (SPAs) has always been spotty. Core Web Vitals and other metrics are based around the concept of full page loads — but, SPAs load once, then swap content dynamically as users "soft" navigate. Everything after that first load has been somewhat invisible.

That changes with soft-navigation attribution, which lets browsers detect when an in-page navigation occurs — making it possible to collect accurate Core Web Vitals for each soft navigation.

Calibre RUM now measures soft navigations automatically, with no additional configuration required. Browser support is currently Chrome 151+, Edge 151+, so bare in mind that current observations will be limited, until browser support improves.

You’ll see Core Web Vitals for your SPA with accurate URL attribution, giving you a much clearer picture of real user experience across your sites and apps.

Filter any report to see soft navigations on their own, or view by navigation type to compare them against the initial load.

If you already run Calibre RUM, there’s nothing to install or turn on: soft navigations start flowing in automatically.

To learn more, see Filtering RUM reports and the RUM documentation, or read about the evolution of soft navigations in Google Chrome’s Monitoring soft navigations explainer.

Until now, managing which page URLs are monitored by Calibre’s Synthetic monitoring was a manual job. From today, Calibre can automatically crawl your site, looking for the most important pages.

You can type one URL into Calibre, and it will continuously crawl a site; finding new pages, and removing old ones, so you can monitor whole websites without having to think about constantly updating settings.

The new Synthetic site scan crawler uses signals such as internal link structure, navigation placement, and other factors to determine which pages are most important. Just like real users would when navigating your site.

Updated Site creation flow#

When you add a Site, you’ll see a new option to Discover pages automatically. Alternatively, you can continue to Add pages manually.

Once your Site has been created, you can watch data come in from Synthetic tests and Google’s CrUX (when available), or install RUM monitoring while you wait.

If you have an existing Site, you can switch to Discover pages automatically from Synthetic → Settings → Pages. After turning on this option, Calibre will scan your Site and monitor pages automatically.

You can follow our getting started guide for a walk-through of the new setup flow, or check the pages documentation for more details.

SAML Single Sign-On is now fully self-service, and available on the Team plan as well as Company.

The new Authentication panel (Organisation settings → Authentication) allows admins to configure SAML SSO and SCIM provisioning.

Domains are verified using DNS TXT records, and the connection is tested to ensure your IdP is configured correctly, avoiding account lockouts.

Once you’ve verified your domain and tested the connection, you can enforce SAML SSO logins for your team, and optionally enable SCIM provisioning to automatically create, update and deactivate accounts from your identity provider.

We’ve also updated our Security and Privacy pages to cover SSO sessions, enforcement and how SSO account data is provisioned.

Read the documentation for more information on how to set up SAML SSO and SCIM provisioning for your organisation.

The Audit log gives a complete, organisation-wide record of security and administrative activity: sign-ins, member and role changes, credential and API key activity, billing and SSO changes, and more. It's built for compliance, security reviews and investigations, giving you a clear audit trail of who did what, when, and from where.

Filter by event type, time range or a single person, export a month of history at a time as NDJSON, or stream every event to your SIEM in real time with signed webhooks.

You can also choose how long events are kept, and place a legal hold to preserve everything during an investigation or for compliance.

The Audit log is available now on the Company Plan. Read the documentation for setup, the event payload and signature verification.

Today we’re shipping a rebuilt interface, designed around an all-new primary navigation.

After a few years of growth and new features, we wanted to consolidate how people move around the app, and make it faster to get to the data that matters.

Navigation is now a single, focused left-hand sidebar. At the top is a switcher: open it from anywhere with a keyboard shortcut (Command-K, or Ctrl-K), start typing, and jump straight to any site. You’ll see a list of the most recent sites, and you can search for others quickly.

The new Sites /home brings all your sites into one comparison view. You can switch between different team workspaces to compare sites by Synthetic, CrUX or RUM.

Across Calibre, you’ll also notice a tighter, more cohesive design: more consistent typography and spacing, a clearer icon set, and a new loading skeleton built to avoid layout shift, so the app settles into place sooner.

We’ve made big improvements to CrUX monitoring, giving you a clearer view of how real users experience your site around the world. It now includes UX Rating, a brand-new Audience report, and a world map on every metric page, each adding a new level of insight.

UX Rating, now available for CrUX#

UX Rating is a single score that summarises user experience quality, based on the percentage of users who experienced good performance for Core Web Vitals and Time to First Byte.

It has been part of RUM reporting for a while, and now it’s in CrUX Monitoring too.

You’ll see UX Rating on your Sites list (to compare all your monitored Sites at a glance), on the Pages report for each Site, and on the new Audience report, which breaks it down by country.

The new CrUX Audience report#

The Audience report is the clearest view yet of where your users have a good experience, and where they don’t. You’ll see a world map of your UX Rating, ranked by country.

There’s also a breakdown by device types and navigation types, so you can gain a better understanding of how people experience your site in different contexts.

Metric pages, restructured#

CrUX metric pages have been restructured to make the data easier to understand and act on.

Across the top, you’ll see the current metric value with colour-coded grading against Google’s thresholds.

For LCP, you’ll see LCP subparts and element type: the components that make up LCP, and the element most commonly responsible for it.

Each metric page also includes a world map, so you can see how performance varies by country.

It’s all live now for every Site tracked in Calibre. To open the new worldview, navigate to Site → CrUX → Audience, or read the documentation.

Following on from Compare Website Speed last week, we’re rolling out another upgrade to our free tools: a rebuilt Core Web Vitals Test.

Here’s a live report for ikea.com to dig into while you read.

Now with UX Rating#

Up front, you’ll see UX Rating: the share of real-world experience on your site that rates as good. Every metric a real Chrome user records is graded, and UX Rating is the percentage of those grades that came back good.

It tells you how much of what your visitors experience is actually good, and gives you a faster read than scanning a wall of metrics. Read more about why we built UX Rating.

It also makes for clean comparisons: see how your UX Rating stacks up against competitors, side by side.

See performance country by country#

UX Rating is also plotted globally, rendered on our custom hexagonal world map, so you can see how user experience varies between the countries your site serves. There’s also map data for LCP, CLS, INP and even TTFB.

Country-level CrUX data requires using Google’s BigQuery database, which is complex and eye-wateringly expensive. We’ve done it; you don’t have to. And then we made it free.

Deeper metric reports#

Dig into Largest Contentful Paint, Cumulative Layout Shift and Interaction to Next Paint with up to 9 months of history, experience distributions, and subpart breakdowns.

Spot a degrading trend in LCP, see where your users are getting stuck on INP, or confirm a recent CLS fix landed cleanly.

Powered by real Chrome user data#

The data comes from Google’s Chrome User Experience Report: the same field dataset used in Calibre’s dedicated CrUX Monitoring product.

Use Core Web Vitals Test to:

• See whether any site passes Google’s Core Web Vitals assessment.
• Get a UX Rating that tells you how much of your visitors’ experience is actually good.
• Compare performance across the countries your site serves.
• Trace regressions in LCP, CLS or INP across 9 months of history.
• Share any report by URL, no sign-up needed.

For continuous monitoring of all your websites (and competitors), use Calibre CrUX Monitoring.

Calibre combines CrUX, Real User Monitoring and Synthetic Monitoring, giving you everything you need to monitor and improve your sites. Start a free trial.

Try Core Web Vitals Test on any site:

Today we’re launching a new free tool: Compare Website Speed.

Drop in up to 5 sites and get a ranked leaderboard built from real Chrome user data, sourced from the Chrome User Experience Report. Every comparison gets a permalink and an auto-generated share card that names the winner.

Use it to:

• See who’s fastest among any group of sites: competitors, alternatives or your own portfolio.
• Compare Largest Contentful Paint, Cumulative Layout Shift and Interaction to Next Paint side-by-side, using real user data from the last 28 days.
• Share the result with a permalink, or save the share card as a PNG.

Free, no sign-up required.

Try Compare Website Speed.

Version 7.0.0 of the Calibre CLI and Node.js API is a major release that brings Real User Metrics and Chrome User Experience Report data to the command line.

Now you can access RUM origin-level and page-level metrics, including live visitor counts, Web Vitals, UX ratings, and historical trends.

# Overall site performance summarised
calibre rum summary --site my-site

# See performance for each page
calibre rum pages --site my-site

# and more…

Similar calibre crux summary, calibre crux urls and calibre crux url commands are available for CrUX data.

❯ calibre crux urls --site maap
✔ CrUX Assessment: Fail
Form factors: desktop 43.7% | phone 55.3% | tablet 1.1%

METRIC                    | P75
Cumulative Layout Shift   | 0.52
First Contentful Paint    | 1.33 sec
Interaction to Next Paint | 187 ms
Largest Contentful Paint  | 1.96 sec
Time to First Byte        | 359 ms
Image Load Delay          | 1.04 sec
Image Load Duration       | 286 ms
Image Render Delay        | 246 ms
Time to First Byte        | 418 ms
Round Trip Time           | 96 ms

Like all other commands, you can use --json to get machine-readable output for use in scripts and code agents.

7.0.0 release highlights#

• calibre rum - access Real User Metrics data from your sites
• calibre crux - access origin-level and URL-level Core Web Vitals from the Chrome User Experience Report, automatically collected by Calibre
• calibre synthetic - existing synthetic monitoring commands are now namespaced under calibre synthetic
• calibre deploy - deploy commands elevated to their own top-level namespace
• calibre site list - now shows which data sources (Synthetic, CrUX, RUM) are active per site
• calibre metric-list --type - filter metrics by data source (synthetic, crux, or rum)
• Node.js API - new Crux and Rum programmatic exports

Deprecated aliases with warnings are provided for all moved commands, so existing scripts will continue to work while you migrate.

Using these new commands, you gain programmatic access to all your sites performance data. You can use this to build custom reports, integrate with other tools, or use code agents to interpret performance data and make recommendations.

See the full changelog on GitHub and CLI command documentation for detailed reference.

Our Test Agents are now using Chrome 147 and Lighthouse 13.1.0 to test your Sites.

New: Baseline audit#

This release introduces a new Baseline audit powered by Google Baseline. The audit detects web platform features in use on your pages and reports how well supported they are across browsers.

Each feature is checked against Baseline compatibility data, giving you a clear view of which features are widely supported, which are newly available, and which may still have limited browser support.

This helps you make informed decisions about which features to use, and where polyfills or fallbacks might still be needed.

We’ve rebuilt the Calibre documentation from the ground up. The new docs are faster to navigate, easier to search and designed for clarity.

What’s new#

• Real User Monitoring documentation: Comprehensive new guides covering Calibre RUM installation, configuration, and data privacy
• Automatic light and dark themes: The docs follow your system preference
• Better navigation: A restructured sidebar and page layout make it easier to find what you need
• All-new search: Faster, higher quality results help you get to the right answer quickly
• Improved readability: Refined typography, spacing and formatting

Of course, the new docs site is faster than ever

Explore the new docs at calibreapp.com/docs.

Understanding how your site performs in the real world is critical to delivering a fast, engaging experience. What actually happens when people visit? Does the page load quickly and smoothly? Without unexpected movement or interaction delays? Which pages are slowest? Are users having a good time?

Today, we’re launching Calibre RUM, which captures Core Web Vitals and key UX signals directly from real browser sessions. Now, you can see exactly where your site is fast, where it’s slow, and who is affected.

Find the pages that matter most#

The Pages Leaderboard ranks your pages by traffic volume. You can sort by any metric (LCP, INP, CLS, or others) to find the pages with the worst user experience. Group pages by URL pattern (like /blog/* or /products/*) to see how vital sections of your site perform.

No guesswork. Start with the pages that affect the most users. Fix them, iterate.

See which elements are slow#

Calibre RUM identifies the specific DOM elements responsible for slow renders (LCP), interaction delays (INP), and layout shifts (CLS). Add custom attribution with data-name or elementtiming attributes so it’s clear which element or image is causing them problem.

Understand your audience#

The Audience Report shows what percentage of visitors had a good, to improve, or poor experience. It segments UX experience by page, page grouping, element, country, device type, browser, and more. You’ll see which areas are affected the most, and what’s worth investigating further.

Privacy-first by design#

Calibre RUM collects no cookies, no personal data, and no IP addresses. Session identifiers live in sessionStorage and are discarded when the tab closes. The script supports signature-based subresource integrity (SSRI) to prevent supply-chain tampering, and you can optionally exclude EEA/EU visitors entirely. Learn about data collection, processing, retention, and more.

Predictable, controllable costs#

You set the session sampling rate and data retention period. Monitor at a level that matches your requirements, and budget. Sample by 15% of user sessions or 100%. Hard stop billing, or flexible, it’s up to you.

Get started in minutes#

RUM is included on Free Trial, Starter, Team, and Company plans.

Navigate to Site → Real User Monitoring → Settings to enable it, or read the installation guide or documentation.

You can also read more about the launch in our blog post.

We’ve updated our Terms of Service, Privacy Policy, Data Protection & Compliance, and Security pages alongside a revised Data Processing Agreement.

These updates reflect the addition of Real User Monitoring (RUM) to Calibre’s product suite.

Because RUM collects performance data from real visitor sessions on your website, our legal and security documentation needed to clearly describe how that data is collected, processed, stored, and protected — and what your responsibilities are as a customer deploying RUM.

We also took the opportunity to trim our sub-processor list (removals only, no new additions), and expanded our security page to reflect the full scope of our technical and organisational safeguards.

If you have questions about any of these changes, or want a copy of the revised Data Processing Agreement, contact us at privacy@calibreapp.com.

Our Test Agents are now using Chrome 146 to test your Sites.

Our Test Agents are now using Chrome 145 to test your Sites.

Figuring out which of your pages appear in Google’s Chrome User Experience Report (CrUX) has always meant manual work. Your site might have thousands of pages, but only a subset shows up in the CrUX dataset. Until now, identifying and tracking which URLs to monitor was on you.

Today, Calibre handles that automatically. Automatic page discovery for CrUX Pages finds the most important URLs on your site and tracks them, so you always have up-to-date, real-user performance data with zero manual setup.

How it works#

Calibre periodically spiders your website to build a map of your most important pages. We look at signals like navigation placement and internal link structure to determine which pages are most important, just like real users would when navigating your site.

Those pages are then matched against Google’s CrUX dataset. Pages with sufficient traffic to appear in CrUX are automatically tracked and kept up to date.

If a page doesn’t receive enough traffic to meet Google’s eligibility thresholds, it won’t appear in CrUX Pages. This is a limitation of the CrUX dataset itself.

What you get#

With automatic CrUX Pages discovery, you can:

• See Core Web Vitals data for your most important pages, without manually auditing your site or guessing which URLs to track
• Stay on top of performance as your site evolves and new pages gain traffic
• Identify SEO risk across pages that matter, ranked by real-user experience
• Filter by device type and customise which metrics appear

Available on all plans#

CrUX Pages with automatic page discovery is available on all Calibre plan tiers. No configuration, or upgrades needed.

View your pages by navigating to Site → CrUX → Pages, or learn more about the CrUX Pages Leaderboard.