Readiness · Public reference
How Readiness scores a site
The score is a static audit result. It is not a prediction that an agent can complete every task.
Site scope
Content sites use content, discovery, access, and page structure checks. API products add applicable API checks. Commerce sites add pricing checks. Protocol-specific checks apply when that protocol is observed. Choose a site type when the automatic choice is wrong.
Core score
Model 27 divides earned core weight by verified core weight. Multiply that ratio by 100 for the score. A failed core check earns zero; a partial result can earn part of its weight. Emerging and bonus-only checks are information and do not change the core score.
Coverage and unknown results
Coverage is verified core weight divided by eligible core weight. Unknown evidence reduces coverage, not the site's earned score. A numeric score requires a readable main page and at least 80% core coverage. A block or challenge page receives no score.
Grades and layers
A is 90–100, B is 80–89, C is 70–79, D is 60–69, and F is below 60. Layer summaries show their own coverage. Fix gains are estimates under the selected model, with other evidence held constant.
Versions and limits
Old captures and evaluations remain stored. A new model creates a separate evaluation. Compare scores under the same model. Source HTML, selected links, and bounded probes cannot prove login, checkout, or general task success. External provider failures appear as limits.