Choose the product surface before cloning
Everyone Can Use English points learners toward several Enjoy products. The supplied README says the web edition is available at enjoy.bot and the browser extension supports YouTube and Netflix. It describes a forthcoming desktop edition as a wrapper and enhancement around the web app. These are project statements, not features verified by our sandbox.
The supplied release page labels v0.7.9 as Latest. That release label does not establish feature parity with the hosted edition or extension. The useful first decision is whether you want to learn with the service or develop the repository's application code.
For a Chinese-speaking learner interested in speech practice, the hosted route is a reasonable first evaluation. For a contributor, the repository needs a separate build, launch, and test check. Our dependency installation cannot answer whether either route works well for learning.
What our recorded run established
In our run on September 27, 2026, at commit 3d799132046993eade5a364ddd1e557906854eda, dependency installation returned exit code 0 after 138.154 seconds. The summary rounds that duration to 138 seconds and records 2,681 dependencies with an installation size of 2,606 MB. These are stored measurements from that run, not a fresh execution or a download-bandwidth measurement.
The repository scan recorded 3,510 files, approximately 62,834 source lines, and 350.8 MB. It also detected workspaces and 11 CI workflow files, with no Dockerfile, Compose file, or top-level tests directory. Workflow-file counts do not prove that CI passed, and a missing top-level tests directory does not mean the project has no tests.
The recorded configuration was 3 CPUs, 8 GB of memory, and image lab-node:22. The image name alone does not identify the runtime. The raw detector recorded Node major 20, while the installation tail shows a Node v20.20.2 release download path.
The installation tail includes native dependency compilation, ONNX Runtime library extraction, Nuxt type generation, and Yarn's completion message with warnings. Those operations happened during dependency setup. We did not verify an installed application binary, launch Enjoy, evaluate the extension, use the hosted service, or measure transcription accuracy, teaching outcomes, or workload performance. No audit result was recorded, so this run supplies no security assessment.
Skipped checks do not mean missing tests
The harness skipped its separate build and test steps with reason "no script". The root package explains the distinction: it provides namespaced commands rather than generic root yarn build and yarn test scripts.
Its commands include enjoy:package, enjoy:test, enjoy:test:main, and enjoy:test:renderer, alongside enjoy:start and enjoy:dev. The declaration of these scripts establishes available entry points, not their success. Our run did not execute them. The supplied extracts also do not establish that they use Playwright or cover five operating-system runners.
The root package requires Node >=20.0.0 and declares Yarn 4.6.0. Before contributing, inspect the workspace instructions and verify the relevant application commands on your target system. Do not treat yarn install as proof that the application starts. The supplied root package redirects the older start entry toward yarn enjoy:start; a bare yarn start is not a root script shown in that extract.
Speech practice has documented limits and service choices
The Chinese settings guide describes local Whisper transcription, cloud speech services, text-to-speech, dictionaries, and local library storage. It describes local Whisper as free and warns that compatibility problems can prevent it from working on some systems. We did not test those capabilities or compatibility claims.
The guide says OpenAI transcription requires the user's own key and that Enjoy AI usage deducts from an account balance. It also describes paid Azure transcription. These are documented service arrangements; they are not a current price comparison or proof that every feature runs locally. The supplied settings extract does not establish the cost of pronunciation assessment when a user supplies an OpenAI key.
Issue #1410 gives a more specific explanation than a general claim of broken recognition. A user reported incomplete assessment of a recording longer than 30 seconds. A collaborator replied that a single pronunciation assessment has a 30-second maximum and recommended sentence-by-sentence practice. That makes recording length a concrete consideration for whole-passage practice. It does not establish recognition accuracy within the limit.
Issue #1416 contains a user's report of an extension translation error and their proposed explanation involving Chrome's downloadable state. We did not reproduce the behavior or independently verify that diagnosis. Treat it as a reported problem to check in an extension trial.
Documentation needs a surface-by-surface check
The supplied main README and detailed settings guide are Chinese. That supports recommending them to readers comfortable with Chinese; it does not establish that all English documentation is absent.
In issue #1409, opened August 10, 2026, a user requests instructions covering web-versus-desktop differences, custom audio import, TTS configuration, feature workflows, and troubleshooting. This is evidence of a specific documentation request, not a complete inventory of missing documentation.
Before committing to Enjoy, check the workflow you actually need on the intended surface. A desktop settings page cannot establish how the current hosted edition handles the same task. Likewise, the repository's dependency footprint should not be presented as the cost of using the hosted service.
How to read the scores
These are editorial evidence ratings for repository adoption, not measured learning outcomes. The rubric is: 1 means a documented blocker for the stated use; 2 means useful evidence with major verification gaps; 3 means a documented path with partial verification; 4 means strong evidence and direct verification for the intended scope; 5 requires broad, repeatable verification with few material gaps. Unknown adoption, compatibility, or test results do not earn a high score.
Setup receives 2/5 because dependencies installed but application build, launch, and tests remain unverified. Documentation receives 3/5 because the supplied settings guide explains concrete configuration choices, while the documentation request identifies questions about product surfaces. Community receives 2/5 because the supplied issues show user reports and a collaborator response, but do not measure adoption, support reliability, or resolution rates. Maturity receives 2/5 because v0.7.9 is a release reference and the README describes a product transition, while our run did not verify application readiness.
Pick a narrower tool when the method is clear
Anki describes itself as a spaced repetition program. Consider it when that is the method you want, without assuming anything here about its mobile clients or card ecosystem.
LinguaCafe describes a self-hosted reading workflow with word lookup and later review. Its README warns that it remains in active development and currently supports one user per server. Consider it when reading and vocabulary review are your main requirements.
For Enjoy, the next useful step is a trial of the exact hosted or extension workflow you need, or a separate verification of the namespaced application commands if you intend to work with the source. Our recorded installation provides a starting point for that evaluation.