Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Email Checker by EntzAI
v1.1.1Automated email assistant for Apple Mail. Runs on a schedule, scores priority, drafts AI replies, and emails you a report. Manage your inbox from Telegram or...
⭐ 0· 108·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
Name/description match the included code: Python scripts use osascript to read Mail.app, score messages, draft replies, and send reports. Required binaries (python3, osascript) and Mail.app are appropriate. However, the code references a get_unread_emails.scpt AppleScript file used to fetch messages but that .scpt is not present in the provided file manifest — if actually missing the checker cannot fetch unread emails as described.
Instruction Scope
SKILL.md and setup.sh instruct the user to grant Terminal permission to control Mail.app (Automation) and possibly Full Disk Access, and to install a cron job. Those are necessary for the stated functionality but are high-privilege actions: the scripts will read mailbox contents and send messages via Mail.app. The README/SKILL.md also expect OpenClaw to edit config/settings.json directly; that gives the agent write access to runtime config. The instructions are fairly specific (not open-ended), but they do expand the agent's operational surface to persistent, privileged automation.
Install Mechanism
No external download/install spec is included; this is instruction-and-source bundled in the skill. The setup is local (setup.sh writes config and installs a crontab). There are no downloads from arbitrary URLs, package manager installs, or extract steps in the skill package — low install-mechanism risk.
Credentials
The registry metadata declares no required env vars. The interactive setup optionally requests an LLM API key (OpenAI or other) and stores it in config/settings.json (gitignored). Requesting API keys is proportionate to the LLM feature. Be aware the key is stored on disk in a file that the skill and any process with file access can read; OpenClaw edits that config as part of its integration, which could expose or change stored credentials.
Persistence & Privilege
setup.sh (the provided installer) will install a cron job to run the checker on a schedule and instruct the user to grant Terminal Automation/Full Disk Access. That creates persistent background activity outside the immediate agent invocation and gives Mail.app automation rights. The skill metadata does not set always:true, but the installed cron plus automation permissions increases the blast radius — review and approve these system-level changes before proceeding.
What to consider before installing
Plain-language recommendations before you install:
- Confirm the missing AppleScript: The Python checker expects a get_unread_emails.scpt file (used to read unread Mail.app messages), but that .scpt is not present in the provided manifest. Ask the author or check the repository for that file; without it the checker will fail.
- Understand the privileges you must grant: setup.sh requires you to allow Terminal to control Mail.app and may ask for Full Disk Access if running from cron. That lets the scripts read and send email on your behalf — only grant this on a machine/account you trust for this purpose.
- Cron persistence: setup.sh installs a crontab entry so the checker runs on a schedule. If you prefer manual runs, decline crontab installation and run the script manually. If you do install cron, note the skill will run even when you are not actively using the agent.
- LLM API keys live on disk: If you select OpenAI or another remote LLM, the API key is stored in config/settings.json. Ensure that file is protected, not in source control, and rotate the key if you stop using the skill. Prefer local LLMs (LM Studio/Ollama) if you don't want keys sent to external APIs.
- Review AppleScript string interpolation and inputs: The code constructs AppleScript snippets using subject/to addresses and other fields. This is expected for AppleScript-based Mail control, but if you allow external inputs (via files or agent commands) ensure they are trusted to avoid injection-like issues.
- Test in an isolated account: Run the setup and test on a disposable macOS account or VM with a dedicated bot inbox (as suggested in README) before giving it access to your primary email.
If you want, I can: (a) search the repo for the missing get_unread_emails.scpt, (b) point out exact lines in the scripts that construct AppleScript or run subprocesses for a closer review, or (c) generate a checklist of settings/permissions to review on macOS before installation.Like a lobster shell, security has layers — review code before you run it.
latestvk97dcyfw035qz2t1dhynvm6w3983hqcg
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
📬 Clawdis
OSmacOS
Binspython3, osascript