Only the latest released version receives security fixes.

Please do not open a public issue. Report it privately through GitHub Security Advisories. Include steps to reproduce and the impact. You should get an answer within 7 days.

tper acts on your behalf towards TPER and Trenitalia, from your computer. Most of it only reads public information. The main risks are:

• Personal data exposure. The complaint profile (name, codice fiscale, address, email) is stored in ~/.config/tper-cli/profile.toml with mode 0600. No command prints it in full (previews and profile show mask values), and it is only sent to TPER's own complaint form. The complaint history stores a hash, the line and the date, never the text or the details. Leaks of these values, in output, logs or errors, are in scope.
• Complaints filed by mistake or by manipulation. An agent might file, repeat or alter a complaint, or be steered by text in a news item. Mitigations are mandatory --yes, --dry-run, duplicate detection, read-only mode, and profile init being interactive-only. Bypasses of these controls are in scope.
• Prompt injection through third-party text. News, notices and stop names are written by others. The CLI returns them as data and the agent skill tells agents never to follow instructions found in them. Terminal escape sequences are stripped from tables.
• Injection and SSRF through arguments (stop codes, lines, places, dates) into URLs or form fields, and redirects to other hosts. Arguments are validated, and the HTTP layer only talks to a fixed allowlist of hosts, including across redirects.
• Hostile XML in feeds (entity expansion). Feeds are parsed with defusedxml.

Out of scope: vulnerabilities in the TPER or Trenitalia services themselves, and a compromised local user account.