What it is
The IBM Equal Access Toolkit is IBM’s open source set of accessibility tools. Its repository holds a browser extension, test integrations for Node, Karma and Cypress, and the rule engine they all share. Supervisor uses the rule engine. The project is published under the Apache License 2.0.
- Project: github.com/IBMa/equal-access
- IBM’s toolkit site: ibm.com/able/toolkit
What it finds
IBM’s engine is a rule-based checker like axe-core, with its own rule set and its own reading of WCAG. It grades what it finds: a violation, a potential violation that a person should look at, or a recommendation. Its rules for ARIA widgets, images and tables overlap with other checkers but are not the same, which is what makes a second opinion useful.
What it cannot find
It reads code and styles, so it shares the limits of every rule-based checker: it cannot judge whether texts make sense, whether a task can be finished with a keyboard, or whether media has good alternatives.
How Supervisor uses it
- The IBM engine runs after axe-core and QualWeb on every page of a crawl, at both sizes.
- Supervisor keeps three of its rules:
aria_widget_labelled,img_alt_nullandtable_headers_related. - The others are left out on purpose. On Supervisor’s test pages with known problems, the rules left out more often pointed at a different WCAG criterion than the one the page broke. A rule that a new version adds stays out until it has been scored.
- Only outright violations are kept. “Potential violation” verdicts are left out.
- A finding is kept only for an element that no earlier checker reported for the same WCAG criterion.
- Every IBM finding waits for a person as “needs review”. None goes into the accessibility statement by itself.
- The engine does not run when a check follows a journey.
In the issue list, IBM findings have rule ids that start with ibm:.
Supervisor is an independent service. It is not affiliated with IBM or the Equal Access project.