Skip to content- Sources: primary
- Summary: Vercel Agent runs as a Vercel App limited to the user's existing permissions, and read-only work uses a baseline token of read: scopes. Plans that need more access list scopes per task, are validated against the same scope registry the OAuth service uses, and on approval yield a token the runtime only uses while the plan is active. Vercel notes that a scope like deployment write is broader than the single redeploy a plan names.
- Why it matters: It is a concrete pattern for giving production agents task-scoped authority without per-action approval prompts, with its stated gaps on scope precision.