Vercel Agent now installs private packages from npm and custom registries
Source
vercel.com
Author
Matan Kushner
Date
Context
Vercel Agent can install private software dependencies using credentials configured as team-shared environment variables. This lets an AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → work with projects that depend on packages unavailable from public registries.
The announcement specifies NPM_tokenThe chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.Full definition → for private npm packages and NPM_RC for custom or multiple registries. These must be shared variables for Development or Preview; project-scoped variables are not read. It also states that credential values remain outside the sandboxAn isolated environment where AI-generated code or agent actions run without being able to touch anything real.Full definition →, allowing package installation without making the underlying secrets readable by the agent.
Key takeaways · AI-distilled
Set NPM_TOKEN as a shared Development or Preview environment variable for private packages on registry.npmjs.org, or NPM_RC to configure custom or multiple registries.
Vercel Agent reads only team-shared variables, not project-scoped ones, and the credential values stay outside the sandbox so the agent itself cannot read them.
npm, pnpm and classic Yarn inside Agent sessions authenticate the same way they do in Vercel builds.
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
sandbox — An isolated environment where AI-generated code or agent actions run without being able to touch anything real.
