# How Vercel Agent accesses production safely

By Steven Van · 2026-10-23

Vercel Agent starts read-only, then asks for approval of a plan that names the scopes it needs, available to all members of Pro and Enterprise teams.

Vercel has published how [Vercel Agent](<https://creatorstoolbox.com/tools/vercel>) gets access to production systems, in [a post on its blog](<https://vercel.com/blog/how-vercel-agent-accesses-production-safely>). The agent investigates a problem first, then proposes a plan that lists what it would do and what access it needs. Approving the plan grants the additional permissions.

## Why the plan sets the permissions

Vercel describes two problems with the alternatives. Broad access means a mistake can affect unrelated resources. Per-step approval means users approve individual actions before they know how the agent intends to solve the problem, and repeated prompts can turn review into a habit of clicking approve. With a plan, the user reviews the proposed fix before authorizing any changes.

## Read-only work runs automatically

For a broken production deployment, the agent might inspect recent deployments and build logs, then compare the failed deployment with the last known-good one. Read-only access lets that investigation run without approval. A short read-only lookup needs no plan. For longer investigations, the agent creates a read-only plan so the user can follow its tasks and progress while it runs.

If the investigation leads to a redeploy, the agent requests the extra access in a plan that can be approved in one decision. Any plan that requests access beyond the read-only baseline waits for approval as a whole, including any read-only tasks it contains.

## Scopes and plan structure

Access is defined by scopes, the named capabilities recognized by the Vercel authorization system. In Vercel's example, the investigation plan requests read:deployment and the redeploy plan requests read-write:deployment.

The post describes a plan as connecting four things:

- **Intent:** the outcome the user asks for.

- **Work:** the tasks the agent proposes to achieve it.

- **Access:** the scopes those tasks require.

- **Enforcement:** the credentials and API checks that enforce the access limits.

Each stored plan records its tasks and progress.

## Availability

Per the [Vercel Agent page](<https://vercel.com/agent>), the agent is available to all members of a team on Pro and Enterprise plans, with no seat license. It can be used in the Vercel dashboard or in Slack by mentioning @Vercel. Pricing is the Vercel Token Rate plus provider inference costs, and approving a plan is billed on demand.

[![](<https://lishhsx6kmthaacj.public.blob.vercel-storage.com/agent-og.png>)](<https://vercel.com/agent>)

![Vercel](<https://www.google.com/s2/favicons?domain=vercel.com&sz=128>)

Vercel

The platform for frontend developers — deploy, preview, and scale web apps and AI agents with zero config.

[View Vercel →](<https://creatorstoolbox.com/tools/vercel>)

Sources

[Original source](<https://vercel.com/blog/how-vercel-agent-accesses-production-safely>)

[Read How Vercel Agent accesses production safely on Creators Toolbox](<https://creatorstoolbox.com/blog/vercel-how-vercel-agent-accesses-production-safely>)

---
Canonical source: https://creatorstoolbox.com/blog/vercel-how-vercel-agent-accesses-production-safely