AI-agent control-surface risk in npm packages.
These packages silently write AI-agent control files during install: agent instructions, MCP server entries, hooks or permission settings for tools such as Claude, Codex or Cursor. This is a dangerous-capability policy finding, not necessarily malware. LPM CLI warns by default and blocks when Firewall is configured to.
Packages
579 packages with LPM's own review, newest scan first. Showing 1–48.
Source confirms an automatic postinstall mutation of Codex’s global control configuration that enables network access for workspace-write sessions. The setting is broader than this packag...
The package automatically alters global Claude settings during npm postinstall to enable private conversation and tool telemetry to a fixed collector. The opt-out does not prevent unconse...
The postinstall hook automatically changes global configuration for detected agent harnesses and bypasses the CLI's selection prompt. This meets the policy for blocking unconsented lifecy...