registry / @expo/agent-cli / 1.0.16
@expo/agent-cli@1.0.16
Agent-native CLI on top of the Expo CLI family
AI Security Review
scanned 17d ago · by lpm-firewall-aiThe manifest has no npm install lifecycle hook, and the flagged agent configuration changes are gated by an explicit setup command and confirmation. The inspected runtime excerpts show ordinary CLI, device, WebSocket, and logging behavior, with no concrete attack identified.
Static reason
One or more suspicious static signals were detected.; previous stored version diff introduced dangerous source
Trigger
A user runs agents:setup and confirms the selected plan, or explicitly passes --yes.
Impact
No confirmed attack surface; the inspected changes are user-requested setup.
Mechanism
Setup can install selected agent skills or plugins and update a managed block in the project’s AGENTS.md. No install-time trigger or unrequested credential or account action was identified.
AI rationale
The manifest has no npm install lifecycle hook, and the flagged agent configuration changes are gated by an explicit setup command and confirmation. The inspected runtime excerpts show ordinary CLI, device, WebSocket, and logging behavior, with no concrete attack identified.
Decision evidence
public snapshotAI called this Clean at 94.0% confidence as Benign with low false-positive risk.
Evidence for AI clean decision
- package.json has no npm install lifecycle hook; its prepublishOnly script runs the clean and production build commands.
- Agent setup requires confirmation unless the user explicitly passes --yes, and setup returns without applying changes when the plan is unconfirmed.
- The managed instruction block is written to the project’s AGENTS.md; setup also exposes this work as an explicit agents:setup command.
- Agent skills use conventional project or user skill directories, and the setup plan selects supported agents and destinations.
- Inspected runtime code handles local device screenshots, WebSocket headers, and local event logging; these excerpts show no unrequested account action or credential transfer.
- Other flagged behavior is ordinary CLI handling: missing Expo is reported with an install suggestion, and device probing determines whether a local device is available.
Evidence against
Behavioral surface
ChildProcessCryptoEnvironmentVarsEvalFilesystem
Source & flagged code
7 flaggedbuild/cli/824.index.jsView file
842const target = (0,windowsShim/* resolveSpawnTarget */.Z)(command.bin, command.args);
L843: const child = spawn(target.command, target.args, {
L844: stdio: ['ignore', fd ?? 'ignore', 'pipe'],
High
Child Process
Package source references child process execution.
build/cli/824.index.jsView on unpkg · L842build/cli/734.index.jsView file
5035L5036: module.exports = eval("require")("bufferutil");
L5037:
Low
Eval
Package source references a known benign dynamic code generation pattern.
build/cli/734.index.jsView on unpkg · L5035build/cli/index.jsView file
1#!/usr/bin/env node
L2: (()=>{var t={6712:(e,t,n)=>{"use strict";e=n.nmd(e);const wrapAnsi16=(e,t)=>(...n)=>{const r=e(...n);return`[${r+t}m`};const wrapAnsi256=(e,t)=>(...n)=>{const r=e(...n);return`[$...
Findings
1 Critical5 High3 Medium6 Low
CriticalPrevious Version Dangerous Deltabuild/cli/458.index.js
HighChild Processbuild/cli/824.index.js
HighSame File Env Network Executionbuild/cli/index.js
HighExternal Ai Agent Control Surface Mutationbuild/cli/248.index.js
HighRuntime Package Installbuild/cli/55.index.js
HighSemantic Analysis Limitedbuild/cli/217.index.js#virtual:normalized:round1
MediumNetwork
MediumEnvironment Vars
MediumStructural Risk Force Deep Review
LowNon Install Lifecycle Scripts
LowScripts Present
LowEvalbuild/cli/734.index.js
LowFilesystem
Affected versions and remediation
This report applies to @expo/agent-cli@1.0.16.
See all versions of @expo/agent-cli for other recorded verdicts.
- Review the evidence and your use of @expo/agent-cli@1.0.16 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.