agentiqa@1.1.89
AI-powered testing for web and mobile apps
AI Security Review
scanned 20d ago · by lpm-firewall-aiSource inspection found a Node version gate and user-configured CLI/provider flows, with no confirmed attack behavior. The scanner's network, gateway, and process-execution signals do not establish an unauthorized transfer or malicious command execution in the inspected excerpts.
Static reason
One or more suspicious static signals were detected.
Trigger
Installation runs the Node version check; CLI network and model requests occur when a user invokes corresponding CLI operations.
Impact
The inspected behavior supports package setup and user-invoked CLI operation; no concrete unauthorized credential transfer or command-output exfiltration was established.
Mechanism
The preinstall hook checks the Node major version. The CLI uses configured provider credentials and API endpoints for model and control-plane requests.
AI rationale
Source inspection found a Node version gate and user-configured CLI/provider flows, with no confirmed attack behavior. The scanner's network, gateway, and process-execution signals do not establish an unauthorized transfer or malicious command execution in the inspected excerpts.
Evidence
package.jsonscripts/preinstall.cjsdist/cli.js
Decision evidence
public snapshotAI called this Clean at 77.0% confidence as Benign with medium false-positive risk.
Evidence for AI clean decision
- The package exposes dist/cli.js as its CLI and its preinstall hook runs scripts/preinstall.cjs.
- The preinstall hook checks the Node version and exits with an upgrade message when it is below 18.
- The CLI selects its control-plane URL from AGENTIQA_API_URL, a saved environment, or its configured default.
- The CLI uses provider keys for an explicitly selected model; OpenAI requests default to api.openai.com, with an optional OPENAI_BASE_URL override.
Evidence against
- The bundled Vercel AI gateway excerpt is provider SDK code that obtains authorization through its configured auth helper; the inspected excerpts do not establish hidden credential redirection or command-output exfiltration.
Behavioral surface
ChildProcessCryptoEnvironmentVarsFilesystemNetworkShell
HighEntropyStringsMinifiedPossibleObfuscationTelemetry
Source & flagged code
9 flaggedpackage.jsonView file
•scripts.preinstall = node scripts/preinstall.cjs
High
Install Time Lifecycle Scripts
Package defines install-time lifecycle scripts.
package.jsonView on unpkg•scripts.preinstall = node scripts/preinstall.cjs
Medium
Ambiguous Install Lifecycle Script
Install-time lifecycle script is not statically allowlisted and needs review.
package.jsonView on unpkgdist/cli.jsView file
5`)+`
L6: `}var tq=new Set(["hint","known-issue","credential"]);function nq(t=Gg){let e=new Set,n=new Set,r=a=>{e.add(a.flag),a.arg&&n.add(a.flag)};for(let a of t.commands)for(let i of a.fla...
L7: Error message: ${ba(e)}`,cause:e}),this[$T]=!0,this.text=t}static isInstance(t){return Me.hasMarker(t,gR)}},yR="AI_LoadAPIKeyError",bR=`vercel.ai.error.${yR}`,uq=Symbol.for(bR),GT,...
High
Child Process
Package source references child process execution.
Findings
9 High4 Medium6 Low
HighInstall Time Lifecycle Scriptspackage.json
HighChild Processdist/cli.js
HighShell
HighSame File Env Network Executiondist/cli.js
HighCredential Redirect Persistencedist/cli.js
HighCommand Output Exfiltrationdist/cli.js
HighRemote Agent Bridgedist/cli.js
HighTrigger Reachable Command Output Exfiltrationdist/cli.js
HighSemantic Analysis Limiteddist/cli.js
MediumAmbiguous Install Lifecycle Scriptpackage.json
MediumNetwork
MediumEnvironment Vars
MediumStructural Risk Force Deep Review
LowScripts Present
Affected versions and remediation
This report applies to agentiqa@1.1.89.
See all versions of agentiqa for other recorded verdicts.
- Review the evidence and your use of agentiqa@1.1.89 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.