Methodology

KLEAR3 exists to answer one question about a project: what does it actually do, and how much of what it says about itself can be checked. Everything below is the process that produces that answer.

What counts as a source

In order: the project's own site, its documentation, its published code and releases, its verifiable announcements, recognised block explorers, official registries, specialised databases, and finally the press for context. A social media post can tell us something is new. It cannot, on its own, establish a team, a funding round, an audit, an integration or a technical claim.

A profile needs at least two relevant sources, one of them primary. Every source is kept with its URL, title, publisher, publication date and the date we consulted it.

One claim, one record

Statements like “the team is public”, “the protocol has been audited” or “the product is usable today” are stored separately, each with its own source, date and confidence level. That is what makes the two blocks on every profile possible: what is verified, and what remains unclear. When sources disagree, the disagreement is shown rather than smoothed over.

What we will not write

• That a project is safe, reliable or risk-free. No verification supports that.
• That a project is a scam, without solid sourcing and careful wording.
• Price targets, forecasts, or anything that reads as buy, sell or staking advice.
• A test we did not run, or a figure without a date.
• A project's own marketing copy, reproduced as if it were a finding.

Where something could not be established, the profile says so. “Not verified” is an honest answer, and it is used often.

Dates, and why they are everywhere

A status is a measurement, not a property. A project marked active three months ago is not an active project, it is an old reading. Every status carries the date it was checked and the date it will be checked again. Emerging or sensitive projects are re-checked weekly, active ones every two to four weeks, durable explainers quarterly.

How automation is used

Discovery, measurement, source collection and the pre-publication checks are automated. Judgement is not: whether a set of sources actually supports a claim is decided by a person before anything is published. Automated drafting is used for structure and rewriting, never as a source, and any claim it produces without a source identifier is rejected rather than published with a hedge.