About ZemDomu
ZemDomu is open-source semantic accessibility static analysis for HTML, JSX, TSX, and Vue source. It checks supported markup and component relationships before the application is rendered. ZemDomu Core is the shared engine behind the CLI, VS Code Extension, and GitHub Action.
Choose your workflow
ZemDomu VS Code Extension
Install the ZemDomu VS Code Extension for semantic accessibility diagnostics and workspace scans across supported HTML, JSX, TSX, and Vue source.
ZemDomu Core and CLI
Install the ZemDomu CLI with npm and run semantic accessibility checks across HTML, JSX, TSX, and Vue files in local workflows.
ZemDomu GitHub Action
Add the ZemDomu GitHub Action to annotate pull requests and fail CI when supported semantic accessibility source findings are reported.
What it can help you find
Local source checks cover supported patterns such as missing image alternatives, unlabeled controls, heading structure, and landmarks. Project-aware analysis follows statically resolvable local React and Vue imports to find supported structural problems that depend on how components are composed.
For example, heading order can change when a page imports a component containing another heading. Supported composed-page checks also use declared page entries to reason about heading and landmark structure. See the cross-component guide for configuration and examples, and the diagnostic reference for source locations and integration formats.
Where static analysis stops
Dynamic component selection, conditional output, runtime state, external components, and slotted content can leave the rendered result unknown. Source diagnostics do not establish correct keyboard behavior, visual contrast, assistive-technology behavior, or WCAG conformance. Use ZemDomu alongside rendered-DOM checks and manual accessibility testing.
The eslint-plugin-jsx-a11y comparison explains the tools' different scopes. The static and runtime testing guide shows how to combine testing layers.
Review the evidence
The cross-component fixture benchmark records paired React and Vue heading-order cases. The diagnostic accuracy corpus publishes reviewed findings, seeded defects, pinned tool versions, and study limitations. These are bounded studies, not guarantees of accuracy or accessibility on every project.