What problem does it solve?
Vercel preview and production deployments protected by Vercel Authentication or Deployment Protection redirect automated browsers to a login page or return 401/403 errors, blocking AI agents from testing them. This Skill authenticates agent-browser sessions using short-lived Trusted Sources OIDC tokens minted from the caller's existing Vercel identity, without disabling protection or exposing static bypass secrets.
Core Features & Use Cases
- OIDC Token Authentication: Mint a development OIDC token with
vc project token and inject it as the x-vercel-trusted-oidc-idp-token header into an agent-browser session without printing or persisting it.
- Failure Diagnosis: Distinguish between protection redirects,
TRUSTED_SOURCES_ENVIRONMENT_MISMATCH errors, and application-level 401/403 responses to route debugging correctly.
- Human Handoff Boundaries: Identify exactly when a human must act, such as adding Trusted Sources rules in the dashboard or upgrading the Vercel CLI past version 53.3.0.
- Use Case: An AI agent needs to verify a protected preview deployment of
my-app.vercel.app. It confirms the Vercel CLI identity, mints a token for the project and scope, opens the URL with the OIDC header, and continues clicking and snapshotting the page in the same authenticated session.
Quick Start
Ask the agent to open your protected Vercel preview URL with agent-browser using a Trusted Sources OIDC token from the Vercel CLI, then take an interactive snapshot of the page.
