AI Security Review
scanned 7d ago · by lpm-firewall-aiNo confirmed malicious attack surface was established. Risky primitives are limited to explicit CLI/dev scaffolding commands and normal frontend API helpers.
Decision evidence
public snapshot- dist/commands/cli.js exposes user-invoked shell commands for npx barrelsby, next, tsc, eslint, and concurrently
- dist/commands/blueprint.js writes generated app/<route>/page.tsx files from local blueprint JSON and stub input
- dist/api/api.js sends requests to caller/env-provided URL or NEXT_PUBLIC_API_HOST with auth bearer
- package.json has no preinstall/install/postinstall lifecycle hooks and no bin entry
- dist/index.js only re-exports library modules and does not run CLI code on import
- No evidence of credential harvesting, hardcoded exfiltration endpoint, persistence, AI-agent surface mutation, eval/vm, or native/binary loading
- Network behavior is normal API client/socket cache functionality using caller configuration
Source & flagged code
4 flagged · loading sourceThis package version adds a dangerous source file absent from the previous stored version; route for source-aware review.
dist/commands/cli.jsView on unpkgPackage source references child process execution.
dist/commands/cli.jsView on unpkg · L6Package source invokes a package manager install command at runtime.
dist/commands/cli.jsView on unpkg · L40