Catching accessibility issues after code ships costs teams time, trust, and rework. accessiBe’s new Code Agent moves that review into the GitHub pull request itself, at the moment code is written.

accessiBe, a global leader in digital accessibility, announced Code Agent. The new capability is live in beta for eligible accessFlow customers. Code Agent reviews GitHub pull requests against WCAG 2.2 AA. It flags the exact line causing an issue and proposes the fix inline. Developers accept it, reject it, or discuss it in the thread. Nothing merges without their review.

Accessibility issues are written in code. But they’ve historically reached engineering from outside the code itself: a ticket, a screenshot, a reference number that isn’t a line of code. By the time it arrives back to the developer, the code has often shipped and the team has moved on to the next sprint.

Code Agent moves that review to where the code already lives: inside the pull request, next to tests, lint, and build, while the author is still in the file and still the one deciding what ships. The fix happens right there, in GitHub, before the change ever reaches production, while the developer is working on the code and they are still the one deciding what ships.

Marketing Technology News: MarTech Interview with Mark Listes, CEO @ Pendulum Intelligence

How it works

Code Agent runs automatically when a pull request opens or updates, and it can also be triggered manually at any time.

It flags the exact line responsible for an issue and writes a suggested fix directly in the pull request, the same way a teammate’s review comment would appear. Teams can set severity thresholds and exclude specific files or rules, so the review stays useful instead of becoming noise. They can also require it to pass before merge, the same way they already gate on tests and lint.

Code Agent supports Vanilla HTML/JavaScript, React, and Next.js today, with framework detection teams can override.

“Accessibility needs to be how software gets built, not something layered on after,” said Robert Lopez, CEO of accessiBe. “Every issue that reaches a developer after code has shipped costs more to fix than it would have the first time, and leaves that exposure sitting in production longer than it should. Code Agent works where engineers are, so an issue gets caught before it ever ships.”

Marketing Technology News: How MarTech Is Enabling Autonomous Brand Engagement Across Channels?

accessiBe’s unified engine, across every stage

Code Agent runs on the same unified accessibility engine that already powers accessiBe’s platform: accessWidget for runtime remediation, accessFlow’s monitoring and remediation tools, and accessScan for automated scanning and audit history. A fix proposed in a pull request reflects the same standard the site is measured against once it’s live.

“We’ve been talking about shifting accessibility left for a long time. Code Agent is what that looks like in practice,” said Sharon Uda, VP of R&D at accessiBe. “It reviews a pull request the same way our engine already reviews a live site, so it narrows the gap with a continuous accessibility standard from the first commit through production. Developers are in control and accept or reject every change themselves, so nothing ships without a human in the loop deciding it’s right.”