npm security reports, with evidence.

This is an unconsented postinstall mutation of a broad AI-agent control surface that enables collection of sensitive Claude session content. The opt-out does not prevent the default autom...

The automatic lifecycle hook mutates a global OpenCode agent configuration and deploys high-capability autonomous agent definitions. This is a concrete unconsented AI-agent control-surfac...

The automatic postinstall hook mutates a foreign global AI-agent toolchain and persistent user configuration. This is concrete unconsented install-time control-surface mutation, so it mee...

OpenSSF Malicious Packages via OSV confirms kartykp-prod-oidc-test-pkg@1.0.3 as malicious (MAL-2026-16197): Malicious code in kartykp-prod-oidc-test-pkg (npm)

OpenSSF Malicious Packages via OSV confirms kartykp-token-pkg@1.0.2 as malicious (MAL-2026-16198): Malicious code in kartykp-token-pkg (npm)

Automatic installation changes global Claude configuration to capture and export sensitive agent-session content. The opt-out environment variable does not establish prior user consent.

This is unconsented install-time mutation of SSH credentials, local SSH configuration, remote authorized keys, system packages, and background processes. The behavior is concrete and reac...

This is an unconsented postinstall mutation of a broad AI-agent control surface that enables remote collection of sensitive AI-session content. The behavior meets the install-hook abuse b...

The automatic postinstall mutates broad AI-agent skill directories for two unrelated agent environments and can remove existing destination content. This is unconsented install-time contr...

This is an unconsented postinstall mutation of a broad AI-agent control surface that enables export of sensitive session content. The opt-out environment variable does not establish prior...

This is an automatic lifecycle chain that executes a remotely obtained opaque binary and broadly installs AI-agent skills. The embedded hash does not make the unreviewable, unconsented ag...

The automatic lifecycle hook modifies global Claude telemetry settings and enables collection of sensitive AI-session content for remote export. This is concrete unconsented AI-agent cont...

This is an automatic install-time mutation of broad, foreign AI-agent control surfaces, which meets the blocking policy. Package-aligned API traffic does not neutralize that persistence b...

The automatic postinstall hook installs a separate global AI-agent package and persists user-environment changes. This meets the install-time foreign AI-agent control-surface mutation blo...

Direct source inspection confirms two automatic lifecycle hooks that exfiltrate local installation metadata to an external Discord webhook. This is concrete malicious install-time data ex...

Automatic install hooks transmit consumer environment identifiers to an unrelated external endpoint. This is concrete unsolicited data exfiltration, so the package should be blocked.

The automatic lifecycle hook mutates broad AI-agent skill directories and installs enabled collection settings without an explicit user setup action. This meets the install-time AI-agent...

The source contains concrete targeted CSS beacons and a cookie-reading JavaScript payload with an external execution beacon. This is malicious browser-side data and credential targeting,...

This is an unconsented postinstall mutation of a broad AI-agent control surface combined with automatic remote collection of Claude interaction content. The opt-out does not establish pri...

The package contains a reachable remote-code loader that executes unverified network content. This is a concrete remote code execution surface despite the absence of install-time hooks.