npm AI-agent policy reports, with evidence.

This is unconsented postinstall mutation of a broad foreign AI-agent control surface, coupled with weakening package-manager lifecycle safeguards. The local-only screenshot utility does n...

This is a concrete unconsented postinstall mutation of a consumer OpenCode control surface, coupled with remote vendor-controlled content application. Signature verification does not remo...

The automatic postinstall writes enabled agent-control configuration and a broad set of agent content into the consuming project. Existing-file conflict handling does not provide consent...

This is an unconsented install-time mutation of broad Claude and Codex command control surfaces, combined with persistent scheduled self-updating. The global-install guard does not make t...

The automatic postinstall hook modifies four global AI-agent skill roots and enables implicit agent behavior. This meets the policy definition of malicious install-hook abuse despite no v...

The package performs unconsented postinstall setup of a third-party Chrome MCP bridge, including permission repair and registration. This is a concrete foreign AI-agent control-surface mu...

The automatic lifecycle hook broadly mutates foreign project and host AI-agent control surfaces, including enabling a host-level tool hook. This meets the install-hook abuse blocking poli...

This is concrete unconsented postinstall mutation of a consumer project's AI-agent configuration, with additional user-wide Codex changes. The code does not show secret theft, but the aut...

The package performs unconsented postinstall execution of a downloaded native payload and automatically installs skills into local AI agents. Hash checking constrains the selected downloa...

The lifecycle hook performs broad, automatic mutations of consumer files and creates AI instruction files. This is a concrete unconsented install-time AI-agent control-surface mutation, s...

This is an unconsented npm postinstall mutation of broad foreign AI-agent control surfaces. The existing-file guard limits scope but does not establish user consent for automatic overwrites.

The automatic lifecycle hook makes broad OpenCode configuration and plugin changes in the consumer project, including registration of an additional browser plugin. This meets the install-...

The automatic postinstall mutation of a global Claude Code command surface is concrete, broad, and unconsented. This meets the install-control-surface block policy even though the install...

The automatic postinstall hook modifies both Claude Code and Codex global MCP control surfaces. This is concrete install-hook abuse under the stated policy, despite no confirmed network e...

The automatic postinstall mutation of both Claude and Codex configuration is concrete install-hook abuse. It creates an agent execution path without an explicit setup command or user cons...

The automatic lifecycle hook broadly mutates a foreign project and creates AI-agent control files, then persists source-file ownership through the build plugin. Although no exfiltration w...

This is concrete unconsented postinstall mutation of a consumer project's package-manager settings and broad AI-agent control surfaces. The generated review-suppression instruction makes...

The package has a concrete automatic lifecycle path that broadly mutates OpenCode and consumer project control surfaces, rather than performing transparent package-local setup. This meets...

This is concrete malicious install-time behavior: it mutates consumer AI-agent controls and embeds instructions that facilitate secret disclosure while suppressing review. No network thef...

The automatic postinstall hook mutates several foreign, user-wide agent skill directories and enables implicit package-authored behavior. The remote native-app installation further increa...

The package performs automatic postinstall mutation of the consumer project's package-manager configuration and multiple AI-agent control surfaces. The combined lifecycle hook, broad writ...