AI Security Review
scanned 3h ago · by lpm-firewall-aiNo confirmed malicious attack surface is established. The package is an explicit QA-generation CLI that invokes a locally configured AI command and writes project test artifacts.
Decision evidence
public snapshot- `dist/core/config.js` executes the user/project-configured AI CLI; setup phases grant Claude `acceptEdits`.
- `dist/core/run-tests.js` runs configured test commands with `shell: true`, but only through explicit `testforge run`.
- `package.json` has no preinstall, install, postinstall, or prepare hook.
- `dist/cli/index.js` routes behavior only after an explicit `testforge` command.
- `dist/core/ai-generator.js`, `seed-generator.js`, and `harness.js` limit Claude generation calls to read-only tools and write returned code only to generated outputs.
- No direct HTTP client, credential transmission, eval/vm use, binary loading, or foreign AI-agent config mutation appears in `dist/`.
- `dist/core/env-template.js` extracts environment-variable names and writes an example file; it does not export their values.
Source & flagged code
4 flagged · loading sourcePackage source references child process execution.
dist/core/discover.jsView on unpkg · L16Package source invokes a package manager install command at runtime.
dist/core/run-tests.jsView on unpkg · L61This package version adds a dangerous source file absent from the previous stored version; route for source-aware review.
dist/core/ai-generator.jsView on unpkg