Episode
658: Automated Love Crunch
- Podcast
- LINUX Unplugged
- Published
- Mar 16, 2026
- Duration seconds
- 3796
- Processing state
processed- Canonical source
- https://linuxunplugged.com/658
Actions
POST https://stenobird.com/v1/public/podcasts/linux-unplugged/episodes/658-automated-love-crunch/transcription-requests
Idempotently request low-priority transcript generation for this episode.GET https://stenobird.com/podcast/linux-unplugged/658-automated-love-crunch.md
Read the agent-friendly Markdown representation of this episode resource.
Summary
The hosts share progress on recent hardware and software hacking projects, ranging from reverse-engineering diesel heater protocols to automating browser tasks with AI. The discussion explores the intersection of low-level hardware control and high-level LLM automation.
Topics
- ESP32
- ESPHome
- Model Context Protocol
- Browser Automation
- Reverse Engineering
- Linux
- AI Agents
- Hardware Hacking
- VPN
- LLM
Highlights
- Hardware Hack: Reverse-engineering the communication protocol of Chinese diesel heaters to enable ESP32/ESPHome control
- Practical takeaway: Using the Model Context Protocol (MCP) to allow AI agents to interact with and control browser sessions via DevTools
- Failure mode: The difficulty of maintaining complex software environments, illustrated by the 'open heart surgery' of migrating to a new host
- Main idea: The rise of specialized, lightweight tools like LLMfit for testing local AI model performance on specific hardware
- Tool recommendation: Using LosslessCut for high-speed, re-encoding-free video and audio trimming
Chapters
1:00Decentralized Networking: An introduction to Nebula, a decentralized VPN built on open-source principles for secure, scalable infrastructure.5:30Diesel Heater Hacking: Discussing the serviceability of Chinese diesel heaters and the potential for ESP32-based automation.9:55Reverse Engineering Protocols: Deep dive into hacking the proprietary communication protocols of Webasto-style heaters using ESPHome.24:55AI Browser Automation: Demonstrating how Chrome's native MCP server allows AI agents to navigate, click, and take screenshots of web pages.38:20Server Migration Challenges: Reflecting on the complexities and risks of migrating large-scale hosting environments.43:25Hardware and Expansion: Exploring the use of older Lenovo machines and the potential for external GPU expansion.57:25Local LLM Benchmarking: Using LLMfit to determine which AI models run most efficiently on specific local hardware configurations.