next/server-action-checks

Dlaczego

A function marked 'use server' and exported is a public HTTP endpoint: anyone can call it with any arguments, whether or not a page shows the button. A check that lives only in the page, the layout or proxy.ts does not protect it, a FormData value can be anything, and whatever the action returns is serialized to the browser.

Poprawka

At the top of each action: read the session (const session = await auth()), refuse when it is missing, and check that the user may change this record (ownership or role). Parse the input with a schema (z.object({...}).safeParse(Object.fromEntries(formData))) and return field errors for useActionState. Return a small result ({ ok: true } or the errors), never a whole database row. Keep database access in a server-only data layer. Review by hand: a tool cannot tell whether the check is the right one.

Przykłady

// bad
'use server'
export async function deletePost(id: string) { await db.post.delete({ where: { id } }) }
// good
'use server'
export async function deletePost(id: unknown) {
  const session = await auth()
  const postId = PostId.parse(id)
  if (!session || !(await canEditPost(session.user, postId))) return { ok: false }
  await db.post.delete({ where: { id: postId } })
  revalidatePath('/posts')
  return { ok: true }
}

Źródła

• Next.js docs, Data Security (Server Actions, authentication, validating client input, return values)
• Next.js docs, Server Actions guide

Przeczytaj w terminalu

npx github:dawidolko/frontend-skills explain next/server-action-checks