npm AI-agent policy reports, with evidence.

This is an unconsented postinstall mutation of multiple external AI-agent control surfaces, combined with execution of an unverified downloaded binary. That is concrete install-hook abuse...

The package has a concrete automatic postinstall mutation of a foreign OAuth callback page. Under the lifecycle control-surface policy, this warrants blocking despite the lack of demonstr...

This is an unconsented postinstall mutation of a foreign AI SDK's sign-in-adjacent control surface. Although the script has no observed exfiltration or network behavior, its automatic dep...

The automatic postinstall hook modifies a foreign OAuth callback asset rather than only package-owned files. Under the install-control-surface policy, this concrete unconsented mutation w...

The package performs an automatic, persistent rewrite of a foreign dependency's OAuth callback UI during postinstall. That concrete tampering outweighs the absence of demonstrated data th...

The package has a concrete automatic postinstall chain that installs and activates a persistent plugin in a separate user agent environment. The absence of observed network exfiltration d...

The package uses postinstall to install and activate a bundled plugin in another application's user-wide plugin directory. That automatic foreign control-surface mutation meets the instal...

The automatic postinstall chain silently changes Claude Code's user-level agent configuration and installs additional MCP capabilities. This meets the blocking policy for unconsented life...

The automatic lifecycle chain makes broad user-level AI-agent configuration changes and MCP registrations. This meets the policy for unconsented postinstall mutation of a foreign AI-agent...

The guarded condition does not supply consent for a new npm lifecycle event. The postinstall-to-installer chain directly mutates global agent control files, meeting the install-control-su...

This is unconsented install-time mutation of a foreign global AI-agent control surface, combined with an automatic remote payload download. The guard requiring an existing Claude director...

The lifecycle hook performs automatic, cross-host AI-agent configuration changes and routes Codex agents through the package relay. The global-install guard does not establish explicit co...

This is an unconsented postinstall mutation of broad, foreign AI-agent control surfaces, with overwrite semantics and related persistence/network behavior. The lifecycle guard on scheduli...

The automatic lifecycle hook modifies a broad consumer AI-agent control surface rather than limiting setup to an explicit command or package-owned location. This meets the install-hook ab...

The automatic lifecycle changes the consumer's OpenCode configuration and activates command-capable MCP integrations without an explicit user setup action. This meets the install-time AI-...

This is a concrete automatic postinstall modification of foreign AI-agent control surfaces, including enabling a Codex MCP plugin. That behavior meets the install-control-surface block po...

The automatic postinstall path broadly modifies foreign project and user-global AI-agent skill directories, including destructive replacement of existing directories. This meets the insta...

The package has a concrete automatic lifecycle path that installs foreign Claude plugins into the consumer project. That exceeds first-party extension setup and meets the policy's install...

This is unconsented install-time persistence for an MCP control service, combined with remote code delivery and a self-reinstallation path. Integrity checks on the downloaded archive do n...

The package has a concrete automatic postinstall chain that modifies Claude Code configuration and installs a privileged agent profile. This meets the install-hook abuse blocking policy.

This is unconsented postinstall mutation of multiple foreign, broad AI-agent control surfaces and the consumer project. The automatic lifecycle chain is sufficient for a publish block.