registry / @doist/comms-cli / 3.4.1

@doist/comms-cli@3.4.1

TypeScript CLI for Comms

AI Security Review

scanned 17d ago · by lpm-firewall-ai

Installing the package automatically rewrites pre-existing AI-agent SKILL.md files in the user home directory. This is an unconsented lifecycle mutation across multiple foreign agent control surfaces.

Static reason

High-risk behavior combination matched malicious policy.; source matched previously finalized malicious package; routed for review

Trigger

npm postinstall during package installation.

Impact

Existing Claude Code, Codex, Cursor, Gemini, Pi, and universal agent skills can be changed without an explicit user command.

Mechanism

The lifecycle entry invokes a global updater, which detects existing configured agent skills and overwrites each with package-provided content.

Attack narrative

The npm postinstall hook loads a global skill updater. It enumerates configured AI-agent directories, detects existing SKILL.md files, and overwrites them with package content. The operation runs automatically on installation and targets several separate agent ecosystems, so it mutates a broad foreign AI-agent control surface without an explicit setup command.

AI rationale

The automatic postinstall path concretely overwrites existing AI-agent skill files across multiple foreign tools. Under the lifecycle control-surface policy, this is blockable install-hook abuse.

Evidence

package.jsondist/postinstall.jsdist/lib/skills/update-installed.jsdist/lib/skills/index.jsdist/lib/skills/create-installer.jsscripts/postinstall.js

@doist/comms-cli@3.4.1

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Affected versions and remediation

Source & flagged code