Agent
semantics-auditor
Audits element choice, roles, names, and structure
@cf/qwen/qwen2.5-coder-32b-instruct
Runs four independent accessibility auditors over the same component source at once - semantics, keyboard, contrast, motion - then reconciles them into one fix list. Sectioning, not voting: each auditor owns a different failure class and none of them sees the others' findings.
Execution
- semantics-auditor + keyboard-auditor + contrast-auditor + motion-auditor · independent parallel reviews
- reconciler
The team
Agent
keyboard-auditor
Audits focus, tab order, and keyboard operation
@cf/qwen/qwen2.5-coder-32b-instruct
Agent
contrast-auditor
Audits color contrast in both light and dark
@cf/openai/gpt-oss-120b
Tools: contrast_ratio
Agent
motion-auditor
Audits animation against reduced-motion and seizure criteria
@cf/openai/gpt-oss-20b
Agent
reconciler
Merges four audits into one ordered fix list
@cf/openai/gpt-oss-120b