AI Security Review
scanned 2h ago · by lpm-firewall-aiLPM blocks this version under the AI-agent control-surface policy. The package runs an install-time `npx forward-nexus sync --pull` command. Source ships broad AI-agent hook examples that alter URL-fetch behavior, but inspected package files do not directly copy them into agent config paths.
Decision evidence
public snapshot- package.json postinstall executes `npx forward-nexus sync --pull` during npm install
- postinstall runs an external package manager command not declared in dependencies
- Package ships agent hook configs for Claude Code, Codex, Cursor, and GitHub Copilot under agents/hooks
- Hook configs deny native web fetch tools and route agents to `npx @taurgis/bonsai`
- agents/hooks/README.md describes hooks as intentionally intrusive
- Lifecycle has CI and FORWARD_NEXUS_SKIP_SYNC opt-out guards
- Shipped hook files are examples per agents/hooks/README.md and are not directly installed by package source inspected
- dist/lib/research/browser.js child_process use only locates/spawns Chrome for user-invoked rendered fetching
- Fetch code includes DNS/private IP checks, redirects limits, content-type checks, and body limits
- No credential harvesting, destructive behavior, or explicit exfiltration found in inspected source
Source & flagged code
5 flagged · loading sourceInstall-time lifecycle script matches a deterministic static-gate block pattern.
package.jsonView on unpkgPackage defines install-time lifecycle scripts.
package.jsonView on unpkgPackage source references child process execution.
dist/lib/research/browser.jsView on unpkg · L1Package source invokes a package manager install command at runtime.
package.json#scripts.postinstallView on unpkg · L1Package ships non-JavaScript build or shell helper files.
agents/hooks/claude-code/hooks/bonsai-web-fetch.shView on unpkg