cleanyourtv Turns Smart TVs Dumb Again

// 1h agoOPENSOURCE RELEASE

cleanyourtv Turns Smart TVs Dumb Again

cleanyourtv uses Claude Code or another coding agent to debloat Android and Google TVs over Wi-Fi, disable ads and telemetry, and install a lightweight Apple TV-style launcher. The open-source workflow requires no root and keeps changes reversible.

// ANALYSIS

cleanyourtv is a compelling example of AI making hardware customization accessible, but it also turns risky system surgery into an approachable afternoon project.

–Uses ADB to inspect devices, disable manufacturer packages, and install a custom launcher
–Supports many Sony, TCL, Xiaomi, Hisense, Philips, and Nvidia Shield models
–Reversible scripts and package backups reduce—but do not eliminate—the risk of breaking core TV functions
–The project highlights growing backlash against smart-TV ads, tracking, and oversized interfaces
–Developers should test model-specific package lists carefully before automating changes

// TAGS

cleanyourtvcoding-agentautomationopen-sourcedevtoollocal-first

DISCOVERED

1h ago

2026-10-19

PUBLISHED

1h ago

2026-10-19

RELEVANCE

6/ 10

AUTHOR

Syntax

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

SECURITY1h ago

Reverse-Engineering MCPs Expose Dangerous Agent Privileges

A source-level review finds that MCP bridges can turn AI-assisted reverse engineering into privileged debugger automation, including memory access, state changes, and execution control. Weak authentication and broad tool surfaces make adversarial binaries and prompt injection especially dangerous.

OPEN SOURCE1h ago

dumbtv strips ads, tracking from smart TVs

dumbtv is an open-source ADB toolkit that debloats Fire TV and Android/Google TV devices without root, disabling telemetry, promotional interfaces, unwanted packages, and ad-heavy launchers. Its changes are largely reversible, though some uninstalled apps cannot be restored automatically.

BENCHMARK1h ago

llm-speed publishes measured local LLM benchmarks

llm-speed is an open-source leaderboard for measured local LLM inference speed across models, hardware, and runtimes. Its signed runs report decode throughput, prefill speed, time to first token, and reproducible configuration details.