Is ai safe?

What ai does

The AI SDK, published on npm as ai, is a TypeScript toolkit for building AI applications and agents. It is provider-agnostic and works with Next.js, React, Svelte, Vue, Angular and Node.js. The npm description calls it the AI SDK by Vercel, and the code lives in vercel/ai under Apache-2.0. It sends prompts to model providers over the network. A plain model string goes through the Vercel AI Gateway by default, while provider packages such as @ai-sdk/openai connect directly. Before installing, decide whether requests should go through the Gateway or straight to a provider, because that choice decides which service receives your prompts and API keys.

No malicious indicators - review capabilities before installing

Notable — review in context (a capability is not a malicious indicator):

Node.js network egress

What to do: Nothing here argues against installing it. Grant the capabilities it lists only if you expect the tool to need them.

ai is an npm package analyzed by SkillTotal's deterministic static scanner. The scan found no malicious indicators. It can: network egress — capabilities are what the code can do, not a verdict on intent. Risk score 0/100 (low).

ai 7.0.173

npm package · npm:ai

LOW

/ 100 risk score

Snapshot · scanned Nov 13, 2026 · ai@7.0.173 · engine 1.1.0 / ruleset 100

Get notified when this component changes

Change alerts aren't live yet. Leave your email and we'll let you know when they launch. We use it only for this and delete it on request.

Automated static-analysis result. It can contain false positives and false negatives, and is not a claim about the intent of ai's authors. Report a false positive.

Capabilities — what this component can do (not a risk score):

network egress

Behavioral traits

How this component maps to the CSA agentic threat model. Descriptive — it never affects the risk score.

Network egress

Interaction & Communication / Direct Communication

Findings (1)

CapabilityNode.js network egressnot scored · severity mediumST-NET-NODE

The component makes outbound network requests.

dist/index.d.ts:6

import { ServerResponse } from "node:http";

dist/index.d.ts:7

import { ServerResponse as ServerResponse$1 } from "http";

dist/index.js:18693

const response = await fetch(api, {

dist/index.js:19630

const response = await fetch(api.token, {

dist/index.js:21369

const response = await fetch(api, {

src/agent/pipe-agent-ui-stream-to-response.ts:7

import type { ServerResponse } from 'node:http';

src/generate-object/stream-object-result.ts:1

import type { ServerResponse } from 'http';

src/generate-object/stream-object.ts:17

import type { ServerResponse } from 'http';

src/generate-text/stream-text-result.ts:9

import type { ServerResponse } from 'node:http';

src/generate-text/stream-text.ts:26

import type { ServerResponse } from 'node:http';

src/realtime/browser-realtime-webrtc.ts:244

const response = await fetch(api, {

src/realtime/realtime-session.ts:365

const response = await fetch(api.token, {

src/text-stream/pipe-text-stream-to-response.ts:1

import type { ServerResponse } from 'node:http';

src/ui-message-stream/pipe-ui-message-stream-to-response.ts:1

import type { ServerResponse } from 'node:http';

src/ui/call-completion-api.ts:66

const response = await fetch(api, {

src/ui/http-chat-transport.ts:204

const response = await fetch(api, {

src/ui/http-chat-transport.ts:280

const response = await fetch(api, {

src/util/write-to-server-response.ts:1

import type { ServerResponse } from 'node:http';

Why it matters: Usually legitimate, but confirm the destinations are expected and no sensitive data leaves.

Fix: Confirm the destination hosts are expected and that no sensitive data is sent off-host.

Maintain ai?

Check the next package before your agent installs it

The free Claude Code plugin scans each npm, PyPI or MCP package with this report's open-source engine before the install command runs. It runs on your machine and needs no account.

Set up the plugin Or scan pull requests in CI with the GitHub Action

Check your own component

Run the same evidence-backed scan on any MCP server, agent skill, or package.

Scan your own component

How we determine this: deterministic static analysis (regex + AST), evidence-anchored, no code execution. Methodology →