Good recommendations
need receipts.

Our job is to help you make a useful choice, with enough evidence to understand the recommendation and its limits.

Three kinds of evidence

1. Documented fact: a statement supported by an official source, with a check date, conditions and scope.
2. Editorial judgment: our interpretation of which project or maintainer a workflow may suit.
3. Hands-on observation: a result from a recorded task, with the plan, inputs, outputs and failures available for review.

This first edition contains documentation reviews. We have not completed hands-on builder tests. No builder profile carries a tested label, star rating or measured performance score.

How we choose a shortlist

We group tools by the work the reader needs to do: publish a website, build an application or retain an independent code workflow. Rankings give an editorial starting order for a specific workflow. They do not measure overall quality or establish a universal winner.

We look at editing, content workflows, code access, hosting, service dependencies and the work after the initial build. A recommendation remains conditional on your requirements.

How facts are maintained

Each key fact has a source, check date and scope. Profiles, directory rows and comparison tables reuse the same data. Unknown support is kept distinct from a confirmed limitation. Conflicting documentation is explained and stays a verification task.

Prices need a currency, billing term, renewal basis and plan entitlement to be comparable. Until those are established, we link to official information and explain what to check. Article update dates change with substantive editorial revisions; they do not automatically become test dates.

What a future test must include

• A common brief and task, including a real interaction and failure state.
• The tool version or date, plan, prompts and revision history.
• A working result, remaining failures and reproducible evidence.
• Published-page checks, mobile behavior and accessibility observations.
• A source-export and clean-build check when portability is evaluated.
• Service dependencies, time and actual charges when making cost claims.

We publish a score only when its method and supporting evidence are ready. A vendor’s product claim does not count as our test result.

Sources and corrections

Follow the source links next to facts and at the end of guides. Use the official page to confirm a changing plan or feature before purchase. Our public fact dataset contains the same scoped facts shown on the website.

Before a correction is published, we verify the updated source and revise affected shared data and editorial conclusions. Send details through our contact and corrections page.

Commercial independence

The first edition contains ordinary official links and no paid placements or affiliate tracking. Future commercial relationships must be disclosed near affected recommendations and must not replace the evidence required for a claim.