npm AI-agent policy reports, with evidence.

The package has a concrete automatic install-time path that mutates project and user-wide AI-agent plugin control surfaces, including adding third-party plugins. This exceeds a transparen...

This is malicious install-hook abuse under the policy because an automatic npm postinstall mutates a consumer’s broad AI-agent control surface and Git hooks. No network theft was needed t...

The automatic postinstall chain performs broad consumer-project agent configuration and permission changes with no explicit user command or consent. This meets the install-control-surface...

The automatic lifecycle hook trusts remote metadata to authorize both a native payload and arbitrary skill content, then executes the payload and installs the content into a global AI ski...

The automatic postinstall chain alters foreign AI-agent configuration and permission state. This is concrete unconsented install-time control-surface mutation.

This is a concrete install-hook abuse chain: an npm postinstall silently installs a separate global AI-agent CLI and persists environment changes. The lack of direct credential theft does...

The package has an automatic postinstall path that writes its instructions into multiple, including discovered, AI-agent skill directories. This is unconsented broad AI-agent control-surf...

This is an automatic postinstall mutation of Claude Code's global or project skill control surface, including overwriting matching skills and executing copied modules. The lack of network...

The automatic postinstall hook writes a globally loaded, behavior-bearing agent skill into multiple existing harnesses. Its broad trigger rules and shell permissions make this a concrete...

The package performs unconsented postinstall mutation of OpenCode configuration and injects binding instructions into future agent sessions. This is concrete install-hook abuse even thoug...

The lifecycle hook performs unconsented broad mutation of consumer AI-agent control files and establishes persistence. This is concrete install-hook abuse, not ordinary package setup.

The package has a concrete, automatic postinstall path that mutates third-party AI-agent control surfaces in the user's home directory. Although no credential theft or network exfiltratio...

This is a typosquatted Grok package whose automatic installer overwrites the shared Grok executable and configuration with an uninspectable binary from a different publisher namespace. Th...

The package combines an automatic lifecycle hook, direct AI-agent instructions to act without a user command, and persistent writes to broad foreign AI-agent control surfaces. This is con...

This package performs broad, unconsented consumer-project mutation during postinstall, including AI-agent control files. That meets the install-hook abuse blocking policy even though no s...

Automatic postinstall writes into global Claude Code and Codex skill registries create an unconsented AI-agent control-surface change. This meets the install-hook abuse blocking policy.

This is an unconsented install-time mutation of broad, foreign AI-agent control surfaces, with persistent instruction content and destructive synchronization. It meets the install-hook ab...

This is a concrete unconsented postinstall agent-integration action performed by an opaque downloaded executable. Digest verification limits replacement risk but does not make the automat...

The package has an automatic install hook that fetches and executes opaque native code and then automatically refreshes AI-agent integrations. Hash verification does not establish the saf...

This is an unconsented postinstall mutation of both a consumer project and an existing user-level AI-agent control surface. The automatic registration is concrete and sufficient for a pub...

The automatic postinstall hook modifies the consuming repository's Claude Code instruction surface. This meets the policy threshold for malicious install-hook abuse even though no network...