Skip to content

Security: hexawulf/PiTasker

Security

SECURITY.md

Security policy

Supported versions

Only the latest release gets security fixes.

Version Supported
latest 2.x yes
< 2.0 no

Reporting a vulnerability

Please don't put details in a public issue. Use GitHub's private reporting (the repository's Security tab → Report a vulnerability) when it is offered; otherwise open an issue that only asks for a private channel, with no details, and you'll get one.

Include the version (the About dialog shows it), what an attacker can do, and the steps to reproduce. The maintainer aims to reply within a week; a fix for a confirmed issue ships as a patch release, credited in the changelog unless you ask otherwise.

Scope

PiTasker edits a real crontab and runs shell commands as its user, so these matter most:

  • authentication or session bypass on any /api route;
  • cross-site request forgery against a state-changing route;
  • a way to write the crontab or start a command without the diff/confirm path;
  • reading files outside LOG_DIR through the logs routes;
  • a fleet agent that answers without a valid token, accepts anything but GET, or sends a secret that redaction should have removed.

Out of scope: attacks that need the operator's own login or shell access, and the content of jobs the operator schedules.

Design notes

  • Every /api route except login and the public auth config needs a session (a test walks the router); sessions are Secure, HttpOnly, SameSite=Lax cookies, regenerated at login.
  • State-changing requests must be same-origin JSON (Origin/Referer and Sec-Fetch-Site checks).
  • Google sign-in accepts only verified emails on FIREBASE_ALLOWED_EMAILS; tokens are verified server-side with revocation checks.
  • Crontab writes go through crontab - without a shell, after a diff, with a backup before and a read-back after.
  • Fleet agents bind to a LAN or tunnel address, store only the SHA-256 of their token, have no request parameters, and run with an empty capability bounding set (no sudo).

There aren't any published security advisories