alt Hacker NewsThe Vercel plugin on Claude Code wants to read your prompts
Comments
> skills are injected into sessions that have nothing to do with Vercel, Next.js, or this plugin's scope
> every skill's trigger rules get evaluated on every prompt and every tool call in every repo, regardless of whether Vercel is in scope
> For users working across multiple projects (some Vercel, some not), this is a fixed ~19k token cost on every session — even when the session is pure backend work, data science, or non-Vercel frontend.
I know everything is vibeslopped nowadays, but how does one even end up shipping something like this? Checking if your plugin/extension/mod works in the contexts you want, and doesn't impact the contexts you don't, seem like the very first step in even creating such a thing. "Where did the engineering go?" feels like too complicated even, where did even thinking the smallest amount go?
The breach of trust here, which is hard to imagine isn't intentional, is enough reason alone to stop using Vercel, and uninstall the plugin. That part is easy. Most of these agents can help you migrate if anything.
The question is on whether these platforms are going to enforce their policies for plugins. For Claude Code in particular this behavior violates their plugin policy (1D) here explicitly: https://support.claude.com/en/articles/13145358-anthropic-so...
It's a really tough problem, but Anthropic is the company I'd bet on to approach this thoughtfully.
The article doesn't link to the code that shows all bash tool uses are sent to Vercel servers by default, i.e. even without opt-in.
Here's the relevant line as a GitHub permalink: https://github.com/vercel/vercel-plugin/blob/b95178c7d8dfb2d...
To be sure, the problem isn't that the plugin injects behavior into the system prompt - that's every plugin and skill, ever.
But this is just such a breach of trust, especially the on-by-default telemetry that includes full bash commands. Per the OOP:
> That middle row. Every bash command - the full command string, not just the tool name - sent to telemetry.vercel.com. File paths, project names, env variable names, infrastructure details. Whatever’s in the command, they get it.
(Needless to say, this is a supply chain attack in every meaningful way, and should be treated as such by security teams.)
And the argument that there's no CLI space to allow for opt-in telemetry is absurd - their readme https://github.com/vercel/vercel-plugin?tab=readme-ov-file#i... literally has you install the Vercel plugin by calling `npx` https://www.npmjs.com/package/plugins which is written by a Vercel employee and could add this opt-in at any time.
IMO Vercel is not a good actor. One could make a good argument that they've embrace-extend-extinguished the entire future of React as an independent and self-contained foundational library, with the complexity of server-side rendering, the undocumented protocols that power it, and the resulting tight coupling to their server environments. Sadly, this behavior doesn't surprise me.
EDIT: That `npx plugins` code? It's not on Github, exists only on NPM, and as of v1.2.9 of that package, if you search https://www.npmjs.com/package/plugins?activeTab=code it literally sends telemetry to https://plugins-telemetry.labs.vercel.dev/t already, on an opt-out basis! I mean, you have to almost admire the confidence.
I use Little Snitch and so far I have only seen Claude Code connect to api.anthropic.com and Sentry for telemetry. I have not seen any Vercel connections, but I always turn off telemetry in everything before I run it. If you log in with OAuth2, it also connects to platform.claude.com . For auto updates, it fetches release info from raw.githubusercontent.com and downloads the actual files from storage.googleapis.com. I think it also uses statsig.anthropic.com for stats. One weird thing, I did see it try to connect to app.nucleus.sh once, and I have no idea why.
Here are some environment variables that you’d like to set, if you’re as paranoid as me:
ANTHROPIC_LOG="debug"
CLAUDE_CODE_ACCOUNT_UUID="11111111-1111-1111-1111-111111111111"
CLAUDE_CODE_DISABLE_ADAPTIVE_THINKING="1"
CLAUDE_CODE_DISABLE_FEEDBACK_SURVEY="1"
CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC="1"
CLAUDE_CODE_DISABLE_TERMINAL_TITLE="1"
CLAUDE_CODE_ENABLE_PROMPT_SUGGESTION="false"
CLAUDE_CODE_ORGANIZATION_UUID="00000000-0000-0000-0000-000000000000"
CLAUDE_CODE_USER_EMAIL="[email protected]"
DISABLE_AUTOUPDATER="1"
DISABLE_ERROR_REPORTING="1"
DISABLE_FEEDBACK_COMMAND="1"
DISABLE_TELEMETRY="1"
ENABLE_CLAUDEAI_MCP_SERVERS="false"
IS_DEMO="1"Engineer at Vercel here who worked on the plugin!
We have been super heads down to the initial versions of the plugin and constantly improving it. Always super happy to hear feedback and track the changes on GitHub. I want to address the notes here:
The plugin is always on, once installed on an agent harness. We do not want to limit to only detected Vecel project, because we also want to help with greenfield projects "Help build me an AI chat app".
We collect the native tool calls and bash commands. These are pipped to our plugin. However, `VERCEL_PLUGIN_TELEMETRY=off` kills all telemetry.
All data is anonymous. We assign a random UUID, but this does not connect back to any personal information or Vercel information.
Prompt telemetry is opt-in and off by default. The hook asks once; if you don't answer, session-end cleanup marks it as disabled. We don't collect prompt text unless you explicitly say yes.
On the consent mechanism: the prompt injection approach is a real constraint of how Claude Code's plugin architecture works today. I mentioned this in the previous GitHub issue - if there's a better approach that surfaces this to users we would love to explore this.
The env var `VERCEL_PLUGIN_TELEMETRY=off` kills all telemetry and keeps the plugin fully functional. We'll make that more visible, and overall make our wording around telemetry more visible for the future.
Overall our goal isn't to only collect data, it's to make the Vercel plugin amazing for building and shipping everything.
The HN headline transformer has mangled this one. The _all your prompts_ part of the original title was important.
@dang
If there were any semblance of liability for software engineering firms things like this wouldn’t happen
Mobile rendering of the post has some issues. Tables are overflowing and not responsive for example
I’ve often seen people say that AI is a multiplier, where a 2x dev becomes a 4x dev, but a -1x dev becomes a -2x dev, etc.
I think it’s fairly easy to tell what impact AI is having at Vercel. Knowing the pre-ai quality of the engineering at that company, I’m not surprised in the AI era they’re pushing stuff like this. I doubt anyone even thought to check it on a repo outside of a Vercel one.
I still use Claude to code in a very stoneage way. I copy c code into the web site/desktop App and type in my prompt. Then i read the output and if i like it i copy paste it into my code. I always felt very old doing that way when things like Claude code exists. Now i fell somehow not so old. All this hacking into my private space using a develpment tool is insane. Also i do not use Git
Oh boy, the guy in the middle wants to take advantage of you! Surprising stuff.
You always had the option to not, ever, touch Vercel.
Every single scam website I have gotten from spam text messages is being hosted on vercel. Not surprising.
This and the comments here make me even more sad that they ended up acquiring the Nuxt project/team :(
“We collect the native tool calls and bash commands”
Holy shit, I cant imagine this to hold for every bash command Claude Code executes. That would be terrible, probably violating GDPR. (The cmd could contain email address etc)
I must be wrong.
We're moving away from Vercel for an increasing number of reasons. But the Vercel brand has fallen so far that we're also moving away from any open source projects they have had any part in creating. The company is almost revolting.
This is a broader pattern I keep seeing with agent plugins/extensions — the permission model is "all or nothing." Once you install a plugin, it gets full context on every session, every prompt.
Compare this to how we think about OAuth scopes or container sandboxing — you'd never ship a CI integration that gets read access to every repo in your org just because it needs to lint one. But that's essentially what's happening here with the token injection across all sessions.
The real problem isn't Vercel specifically, it's that Claude Code's plugin architecture doesn't have granular activation scopes yet. Plugins should declare which project types they apply to and only activate in matching contexts. Until that exists, every plugin author is going to make this same mistake — or exploit it.
AI tools right now remind me of the old days of single-user PC/Mac operating systems without protected memory or preemptive multitasking. You could read any file, write directly to video memory, load machine code into the heap and then jump to it, etc.