registry / linkgravity / 1.5.1

linkgravity@1.5.1

Discord/Telegram bot bridge for the Antigravity (agy) CLI, with voice interaction support

AI Security Review

scanned 15d ago · by lpm-firewall-ai

LPM flags this version as an AI-agent control-surface risk. Installation silently mutates Antigravity's shared hook configuration. The installed wildcard hook can mediate agy tool calls and the Stop hook can keep agent turns active.

Static reason

One or more suspicious static signals were detected.

Trigger

npm postinstall during package installation

Impact

Unconsented persistence in a foreign AI-agent control surface and altered agent execution behavior.

Mechanism

writes broad agy PreToolUse and Stop hooks

Policy narrative

Installing the package executes postinstall, which creates a persistent environment and automatically registers named wildcard PreToolUse and Stop hooks in the user's shared ~/.gemini/config/hooks.json. These hooks execute package-controlled Python for agy events; the runtime also launches agy with permissions skipped. This is an unconsented install-time mutation of a foreign, broad AI-agent control surface.

Rationale

The lifecycle behavior is concrete source code, not a scanner inference. It meets the firewall block policy for unconsented postinstall mutation of a foreign/broad AI-agent control surface.

Evidence

package.jsonnpm-scripts/postinstall.jsnpm-scripts/register-hook.jshooks/hook.pyhooks/stop_hook.pysrc/core/agy_runner.py~/.gemini/config/hooks.json~/.gemini/config/hooks.json.bak~/.gemini/linkgravity/venv

Network endpoints1

linkgravity@1.5.1

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Source & flagged code