npm detections from LPM Firewall.

The source implements a concrete credential-exfiltration path to a non-Facebook endpoint and an enabled-by-default runtime updater that changes the consumer project. These behaviors occur...

The package performs unconsented postinstall mutation of a foreign AI-agent configuration and registers an executable server that accesses local credentials. This meets the install-hook a...

Static inspection establishes automatic, project-wide configuration and AI-agent control-surface writes during installation. The reviewer-directed instruction to suppress secret-leak find...

The import-time hidden downloader and executable payload runner are concrete malicious behavior. The absence of an install hook does not remove the import-time execution risk.

This is an unconsented postinstall mutation of a broad AI-agent control surface. The automatic configuration change makes package-authored instructions load in later OpenCode sessions.

The automatic installer both provisions OpenCode MCP configuration and applies remotely delivered content into the consumer project. This is a concrete unconsented postinstall AI-agent co...

This is an unconsented postinstall mutation of a consumer AI-agent control surface, with enabled npx-backed tool commands. It meets the install-control-surface block policy despite the ab...

The automatic global postinstall mutates foreign AI-agent configuration and persists a provider and child agents. The global-install guard does not establish explicit consent for these ch...

The automatic postinstall hook silently writes consumer project AI-agent configuration and patches a third-party dependency. This meets the policy threshold for malicious install-hook abu...

The package's postinstall makes broad, persistent changes to a consuming project's agent, package lifecycle, and Git execution configuration. This is concrete unconsented install-time con...

Direct inspection confirms automatic postinstall mutation of foreign AI-agent configuration and agent files. The global-only guard limits scope but does not provide separate consent for t...

The package contains a concrete automatic install-time chain that modifies consumer lifecycle scripts and AI-agent control surfaces, then persists and replays those changes. This meets th...

Hardcoded receiver-controlled webhooks are enabled by default and receive serialized application payloads. This is concrete runtime data exfiltration despite the absence of an install hook.

The reviewed source establishes an automatic postinstall chain that mutates a foreign project and user-wide AI-agent configuration, with detached persistence of its changes. That concrete...

The package's automatic postinstall both changes dependency-script policy and installs persistent AI-agent instructions into the consumer project. Its generated review exclusion conceals...

This is unconsented postinstall mutation of broad, foreign AI-agent control surfaces, combined with instructions that autonomously redirect agent work and a self-update path that repeats...

The automatic postinstall hook globally installs a third-party AI-agent CLI and persists user-environment changes. That concrete install-time behavior meets the blocking policy despite no...

The automatic postinstall hook broadly modifies the consumer repository’s AI-agent control surface and configures a package-owned MCP command. This meets the policy for malicious install-...

This package has a concrete, automatic postinstall chain that overwrites a consumer Claude skill with broad activation instructions. Under the install-control-surface policy, that is mali...

The package contains a concrete remote collection path for an exam credential and user shell input, activated as part of its exam workflow. Its absence of install hooks does not mitigate...

The lifecycle hook makes broad consumer-project and agent-plugin changes automatically. This meets the blocking policy for unconsented postinstall mutation of a foreign AI-agent control s...