Run Accent Testing

Set up and launch a Coval accent testing sweep end-to-end: one voice agent, one test set, an accent persona pack, shared metrics, a per-persona comparison table built from the results, and a saved multi-run report grouped by Persona (created for you via the API). Mirrors the public cookbook at https://docs.coval.dev/guides/testing-across-accents.

Unlike the audio-quality pack, the accent personas are not built-in. This skill creates them in the user's organization, one per accent voice, and makes each one mirror the org's Standard Customer behavior so the only variable between runs is the speaker's accent.

When the runs complete this skill hands off to analyze-accent-report for the recommended-fixes write-up.

When To Use