All evals
Mozilla

Eval directory

Evals for Mozilla

Eval coverage for Mozilla, mapped from its public product surface.

About Mozilla

Mozilla is a nonprofit-led organization that builds privacy-oriented internet products, led by the Firefox browser. Its product line also includes Firefox Focus, Firefox Relay, Mozilla Monitor, Mozilla VPN, MDN Plus, Thunderbird, and Solo. The pages position these products as putting people ahead of profit, with privacy protections such as tracking blocking, email masking, breach scanning, and encrypted connections.

Industry

privacy-focused consumer internet products (browser, VPN, breach monitoring)

Use the eval library for Mozilla

We'll build out the full library — runnable test cases with inputs, expected behavior, and pass/fail checks — in your Corsac workspace.

Generate your own →

Coverage map

What would you measure for Mozilla?

6 scoring areas · 24 capabilities mapped · grounded in 8 cited pages

Every eval set is graded on

  • Adversarial robustness
  • Workflow quality
  • Safety gates
  • Operator quality

Pass/Fail + LLM judge 1–5 · critical severity flags · negative controls

01

Product portfolio discovery and routing

Whether the surface correctly identifies which Mozilla products exist, what each one is for, and which product answers a given user need — without conflating siblings such as Firefox versus Firefox Focus, or Relay versus Monitor.

Mapped capabilities

4 capabilities

  • Product inventory and one-line purpose

    Firefox, Firefox Focus, Firefox Relay, Mozilla Monitor, Mozilla VPN, MDN Plus, Thunderbird, Solo as listed on the products page.

  • Need-to-product routing

    Mapping a stated goal (mask an email address, check for a breach, build a small business site) to the single correct product.

  • Sibling product disambiguation

    Distinguishing Firefox from the dedicated Firefox Focus privacy browser, and email masking (Relay) from breach monitoring (Monitor).

  • Organizational framing

    Mozilla as a nonprofit-led, mission-driven organization and the Mozilla Foundation donation path, stated without embellishment.

02

Firefox download and platform eligibility

The download surface has real conditional logic: unsupported operating systems must be routed to Extended Support Release rather than the standard build. This is the clearest deterministic behavior on the pages supplied.

Enhanced Tracking Protection comes standard in every Firefox browser. www.mozilla.org

Mapped capabilities

4 capabilities

  • ESR routing on unsupported Windows

    Windows 8.1 and below directs to Firefox ESR, with 64-bit and 32-bit choices.

  • ESR routing on unsupported macOS

    macOS 10.14 and below directs to Firefox ESR.

  • Standard build versus alternate build selection

    Default download path and the 'download a different build' escape hatch.

  • Privacy notice adjacency

    Surfacing the Firefox Privacy Notice alongside the download offer.

Illustrative example

Input
I'm on a MacBook running macOS 10.13 High Sierra and I want to install Firefox. Which download should I use?
Expected behavior
Explains that current Firefox no longer supports macOS 10.14 and below, and directs the user to Firefox ESR (Extended Support Release) as the supported download instead of the standard build.

03

Mozilla VPN plan and feature facts

A purchase-decision surface with specific, checkable claims. Errors here are commercially material, so fidelity to stated numbers and conditions matters more than fluency.

Mapped capabilities

4 capabilities

  • Pricing and term

    50% savings with an annual subscription, $59.88 total plus tax, versus a monthly option.

  • Coverage limits

    Up to 5 connected devices; more than 500 servers across 30+ countries.

  • Refund guarantee conditions

    30-day money-back guarantee, first-time customers only.

  • Feature and protection claims

    No logging or sharing of network data, no bandwidth throttling, multi-hop routing, whole-device protection, open source, Firefox extension for per-site settings (Windows only).

Illustrative example

Input
If I subscribe to Mozilla VPN and don't like it, can I get my money back? I had a subscription a couple of years ago.
Expected behavior
States the 30-day money-back guarantee and notes it applies to first-time customers only, so a returning subscriber should not assume eligibility; does not invent a different refund window or an appeals process.

04

Mozilla Monitor breach exposure workflow

An end-to-end user workflow: scan for exposures, receive breach notifications, and act on leaked personal data. Covers both the mechanics and the limits of what the pages claim.

Get notified instantly if your data is detected in a breach. www.mozilla.org

Mapped capabilities

4 capabilities

  • Scan entry and expected inputs

    Initiating an exposure scan from 'Scan for Exposures' / 'Scan Now' entry points.

  • Exposure categories

    Credit card information, phone number, and other personal data described as breach-detectable.

  • Ongoing monitoring and alerts

    Instant notification when data is detected in a new breach; continual identity monitoring.

  • Data removal framing

    Automatic retrieval of private info and protection from data brokers, described without overstating guarantees.

06

Localization and global delivery

mozilla.org ships in a very large locale set, and localized homepages surface partially different product lineups. Correct locale handling — including right-to-left rendering — is a visible quality signal.

Mapped capabilities

4 capabilities

  • Locale coverage and switching

    The published locale list, including regional variants such as English (US/British/Canadian) and Spanish (Spain/Mexico/Argentina/Chile).

  • Right-to-left locale handling

    Arabic and Hebrew page rendering and directionality.

  • Locale-specific product lineups

    Localized homepages listing additional entries such as 0DIN and Tabstack alongside the core products.

  • Consent surface localization

    Cookie banner text translated, or falling back to English, per locale.

Coverage is mapped from Mozilla's public pages (8 crawled). Examples are illustrative, not real test cases. The runnable eval library — graded inputs, expected behavior, and pass/fail checks — is built when you request it above.

Frequently asked questions

What do the Corsac evals for Mozilla test?+

The coverage map is generated from Mozilla's own public product surface (privacy-focused consumer internet products (browser, VPN, breach monitoring)): 6 scoring areas — Product portfolio discovery and routing, Firefox download and platform eligibility, and Mozilla VPN plan and feature facts, and more — spanning 24 mapped capabilities, each graded on adversarial robustness, workflow quality, safety gates, and operator quality once the library is built.

How are the Mozilla evals scored?+

Every case generated for Mozilla — across Product portfolio discovery and routing and Firefox download and platform eligibility and the other mapped areas — is graded with pass/fail checks plus an LLM judge scoring 1–5 against its expected behavior, with critical-severity flags and negative controls. Only judge-passed evals are published.

How many test cases does the Mozilla library include?+

The full Mozilla library is built on request. The coverage map spans 6 areas and 24 capabilities (for example, Product inventory and one-line purpose and Need-to-product routing under Product portfolio discovery and routing); each becomes graded test cases — inputs, expected behavior, pass/fail checks — in your Corsac workspace.

How do I run these evals against Mozilla or my own agent?+

Request the library with your work email above. We'll build out all 6 mapped Mozilla areas and set them up in a Corsac workspace, where you can run every test case against Mozilla or your own agent with your own data.