Skip to content

Security: silent-diffusion/YALTI-Prompter

Security

SECURITY.md

Security and privacy

Reporting a vulnerability

Please report security issues privately using GitHub's private vulnerability reporting rather than a public issue. Include the YALTI version, Windows version and steps to reproduce. We aim to acknowledge reports within a week and will credit you in the release notes if you wish.

Only the latest release receives security fixes.

Privacy design

  • Speech recognition runs locally in a separate process. Audio is held in memory only for recognition and is never stored or transmitted.
  • Scripts are read from disk and never uploaded.
  • The app makes no network requests during normal use: no telemetry, analytics, crash reporting or accounts. The one exception is updating, which only happens when the user checks for updates or turns on automatic updates (off by default): the app asks the GitHub API for the latest release of this repository and downloads its files from github.com/<repo>/releases/download/ only. A download is used only if its SHA-256 matches the release's SHA256SUMS.txt (or the SHA-512 in electron-builder's latest.yml); releases without checksums are refused.
  • Settings, recent file paths and reading positions are stored in plain JSON in the user's application-data folder (or next to the portable app).

Hardening

  • All windows use context isolation and the Chromium sandbox, with no Node.js access in pages.
  • Pages are served from a private app:// scheme limited to the app's own files, with a strict Content Security Policy.
  • The main process accepts IPC only from the app's own pages and validates every argument.
  • Navigation, new windows and <webview> are blocked.
  • Microphone permission is granted only to the app's own origin; all other permission requests are denied.
  • Markdown is converted to plain styled text; HTML in scripts is never rendered.

There aren't any published security advisories