Proton Mail

Manage ProtonMail emails via Playwright browser automation. Login, read, send, and manage your encrypted inbox.

This skill appears to do what it says (browser automation for ProtonMail) but exercise caution before installing or running it. Items to consider: 1) Manifest mismatch — the SKILL.md asks you to supply PROTON_EMAIL and PROTON_PASSWORD but the registry metadata does not declare these as required credentials; ask the publisher to correct the manifest so you clearly know what secrets will be used. 2) Bot-evasion code — the instructions explicitly disable automation detection and recommend disabling the Chromium sandbox; this can violate ProtonMail's terms and raises operational and security risks (especially the --no-sandbox flag on multi-user hosts). 3) Credential safety — only run this in a trusted, isolated environment (not on shared servers) and prefer ephemeral accounts or Proton-approved methods (Bridge/API) if available. 4) Source verification — the skill’s source is unknown; verify the author/publisher and review the exact code you will run. 5) If you decide to run it, avoid running with --no-sandbox on production/shared machines, ensure the environment variables are injected securely, and consider manual review or running in a locked-down container/VM. If you want, I can draft a checklist of safe operational steps or a safer alternative approach (e.g., Proton Bridge or official APIs) based on your environment.

Type: OpenClaw Skill Name: protonmail-claw Version: 1.0.1 The skill is designed for ProtonMail automation using Playwright, which is a legitimate purpose. However, it includes instructions for installing system dependencies using `sudo apt-get install` and launches Chromium with `--no-sandbox` and `--disable-setuid-sandbox` flags. While these might be necessary for certain environments or to bypass bot detection, they significantly weaken the security posture of the host system by disabling critical browser sandboxing and requiring elevated privileges for installation. These are serious vulnerabilities that could be exploited, classifying the skill as suspicious rather than benign, despite no direct evidence of intentional malicious data exfiltration or unauthorized remote control in the provided code snippets or instructions.