Personal Finish Notifier

ReviewAudited by ClawScan on May 10, 2026.

Overview

The notifier mostly does what it says, but its setup script persistently rewrites Claude Code completion hooks broadly and may overwrite existing hooks.

Before installing, back up ~/.claude/settings.json and check whether you already have Stop or TaskCompleted hooks. Only run the installer if you are comfortable with a persistent completion hook and configure the WhatsApp target, preferably with the self-target guard.

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.

ConcernHigh Confidence

ASI02: Tool Misuse and Exploitation

What this means

Running setup could disable or replace existing Claude Code completion hooks and make this notifier run for every matching completion event.

Why it was flagged

The installer broadly registers the hook for all matches and unconditionally replaces the existing Stop and TaskCompleted hook entries instead of preserving or merging them.

Skill content

entry = [{"matcher": "*", "hooks": [{"type": "command", "command": f"bash {hook_path}", "timeout": 15}]}]
data["Stop"] = entry
data["TaskCompleted"] = entry

Recommendation

Back up ~/.claude/settings.json before installing, review the generated hook entries, and prefer an installer that merges with existing hooks, asks before overwriting, and provides an uninstall path.

NoteHigh Confidence

ASI10: Rogue Agents

What this means

The notifier will keep running on future Claude Code completion events until the hook and settings entry are removed.

Why it was flagged

The setup copies an executable notifier into Claude's hooks directory, creating persistent behavior that is purpose-aligned but remains active after installation.

Skill content

HOOK_PATH="${HOME}/.claude/hooks/mac-task-notify.sh"
cp "${SKILL_DIR}/scripts/notify.sh" "$HOOK_PATH"
chmod +x "$HOOK_PATH"

Recommendation

Install only if you want persistent completion notifications, and document or keep an uninstall procedure for removing the hook and related settings.

NoteHigh Confidence

ASI03: Identity and Privilege Abuse

What this means

A configured OpenClaw/WhatsApp account may send automatic 'Claude has finished.' messages after Claude events.

Why it was flagged

The notifier uses the user's OpenClaw messaging authority to send an outbound WhatsApp message to the configured target, which is expected for the stated purpose.

Skill content

openclaw message send --channel "$channel" --target "$target" --message "${MESSAGE}" --json

Recommendation

Set OPENCLAW_NOTIFY_TARGET carefully, use OPENCLAW_NOTIFY_SELF_TARGET if this should only notify you, and set MAC_NOTIFY_MODE=off to disable delivery.

NoteHigh Confidence

ASI04: Agentic Supply Chain Vulnerabilities

What this means

Pre-install checks may not warn you about the needed CLI tools or the local Claude configuration files that will be changed.

Why it was flagged

The supplied scripts call python3/openclaw and read/write ~/.claude files, so the registry metadata under-declares prerequisites and configuration touchpoints.

Skill content

Required binaries (all must exist): none ... Required config paths: none

Recommendation

Verify python3 and the intended openclaw CLI on PATH before running the scripts, and declare these prerequisites in the skill metadata.