CleanYourTV: AI-Driven Android TV Debloating Meets a 523 KB Apple TV-Style Launcher
CleanYourTV is a working blueprint for using an AI coding assistant to operate real hardware safely. It shows how to constrain an AI's actions with scripts, checks, and reversible records, and the project's careful handling of edge cases—like home screen protection, network ADB insecurity, and firmware drift—sets a high bar for any AI-driven device management tool.
CleanYourTV is an open-source project that combines an AI coding assistant—Claude Code, Codex, or Gemini CLI—with a set of shell scripts and a prompt file to automatically debloat and re-skin Android and Google TVs. The assistant discovers which preinstalled apps exist on the TV, asks for confirmation, and then runs `pm disable-user` to disable telemetry, ad services, and bloatware, uninstalling nothing. It then builds and installs a tiny native launcher (523 KB, pure Java, no network permission) that replaces the home screen with an Apple TV-like interface. The entire process is reversible: backup records are written before any change, and `tools/restore.sh` reverts everything with one command. The project gained 38,000 stars in days, driven by a viral tweet from creator Gathin23, and now includes community-contributed profiles for 316 TV models across 89 brands, a companion Android app, and extensive documentation on safety and limitations. The design emphasizes a strict boundary: the AI handles discovery and proposal, while all destructive actions live in well-tested scripts with home-screen protection, drift detection, and rollback logic.
By separating discovery (AI) from execution (hardcoded scripts), the project limits what an AI can break; any destructive change passes through tested, auditable code.
The launcher's total absence of network permission makes its privacy claim checkable at a glance, a deliberate trade-off that many similar tools avoid.
The depth of edge-case handling—setup wizard dependencies, drift after firmware updates, and the security cost of always-on network ADB—shows that even a 'simple' debloat tool faces real system integration complexity.
The rapid community growth (200+ contributors in four days) signals strong user demand for safe, reversible ways to control Android TV, likely reflecting widespread frustration with ad-heavy default interfaces.