Agent Notify

PassAudited by ClawScan on May 1, 2026.

Overview

Agent Notify appears purpose-aligned for local notification setup, but users should notice that it runs local setup commands, persists agent hooks, and has incomplete Windows/provenance details.

This skill looks appropriate if you want local sound/desktop notifications for an AI coding agent. Before installing, confirm you are comfortable with it modifying agent notification configuration, enabling future hooks, and using the provided macOS/Linux scripts; Windows users should verify the missing Windows script separately.

Findings (4)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

Setup may run local commands and inspect common agent configuration directories on your machine.

Why it was flagged

The setup flow directs local shell commands to detect the OS and agent directories. This is expected for installing notification hooks, but it is still local command execution.

Skill content
Run this command to detect the operating system:
```bash
uname -s 2>/dev/null || echo "Windows"
```
Recommendation

Only choose setup when you intend to install it, and review the commands before allowing them to run.

NoteHigh Confidence
ASI10: Rogue Agents
What this means

After installation, your agent may continue playing sounds or showing notifications when configured events occur.

Why it was flagged

The default configuration enables persistent notification behavior for future agent confirmation and completion events. This is aligned with the skill purpose, but it changes ongoing agent behavior.

Skill content
"Notification": { "enabled": true, "type": "confirm" },
"Stop": { "enabled": true, "type": "done" }
Recommendation

Keep only the hook events you want enabled and verify how to disable or remove the configuration if you no longer need it.

What this means

If you follow the README install command, you may install whatever is currently on that remote branch.

Why it was flagged

The README suggests installing from an unpinned GitHub clone. That may fetch code different from the reviewed registry artifact.

Skill content
git clone https://github.com/Miluer-tcq/agent-notify.git
cp -r agent-notify ~/.claude/skills/
Recommendation

Prefer the reviewed package or a pinned, trusted commit when installing manually from GitHub.

What this means

Windows users may expect complete Windows support even though the referenced Windows implementation is not present in the reviewed artifact set.

Why it was flagged

The README references a Windows notification script, but the provided manifest/code files only include Linux and macOS scripts. This is a transparency/reliability mismatch rather than evidence of malicious behavior.

Skill content
+-- scripts/
|   +-- notify-windows.ps1   # Windows notification
Recommendation

Verify the Windows script exists and review it before using this skill on Windows.