registry / @wowok/skills / 2.2.1

@wowok/skills@2.2.1

WoWok AI Skills for Claude and other AI assistants - Dialogue orchestration layer on top of the WoWok MCP server (rules/reference knowledge is served by MCP directly since v2.0.0)

AI Security Review

scanned 10h ago · by lpm-firewall-ai

LPM flags this version as an AI-agent control-surface risk. Installing the package silently provisions a Wowok MCP server across multiple unrelated AI-agent clients. It creates or modifies global agent configuration and launches a registry-fetched MCP package.

Static reason

High-risk behavior combination matched malicious policy.

Trigger

npm postinstall

Impact

Adds an npx @wowok/agent-mcp launch entry to user-level AI-client configuration and may restart its process.

Mechanism

broad AI-agent configuration mutation plus global MCP installation

Policy narrative

On npm postinstall, the package defaults to eleven AI-client targets, copies skills, installs or upgrades @wowok/agent-mcp globally, and registers it through user-level MCP configuration. The inserted entries invoke npx -y @wowok/agent-mcp, allowing later registry resolution and execution by the affected clients. It then may terminate an existing Wowok MCP process to force a restart. This is unconsented mutation of a broad, foreign AI-agent control surface.

Rationale

The lifecycle script performs the concrete broad agent-control-surface mutation that the firewall policy reserves for blocking. Although the configuration is branded Wowok and no direct credential exfiltration appears in this package, automatic postinstall action across unrelated clients remains disqualifying.

Evidence

package.jsonscripts/install.js~/.claude/settings.json~/.codex/config.toml~/.copilot/mcp-config.json~/.codeium/windsurf/mcp_config.json~/.roo/mcp_settings.json~/.kilocode/cli/global/settings/mcp_settings.json

@wowok/skills@2.2.1

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Source & flagged code