Show HN: EchoKit – An open-source, ESP32-based AI voice agent with a Rust server (www.instructables.com)

EchoKit is an open-source project that turns an ESP32-S3 dev board plus an audio extension, mini speaker and 1.54" LCD into a talking AI agent. The tutorial walks you through quick assembly, flashing firmware (via the ESP32 Launchpad GUI or espflash CLI), and pairing the device over Bluetooth in a browser to provision 2.4 GHz Wi‑Fi and a server WebSocket URL. You can use EchoKit’s public hubs (ws://indie.echokit.dev/ws/ or ws://hk.echokit.dev/ws/) for a fast start or run your own Rust-based server for full control. Once connected, pressing the device button enters chat mode and the device runs a simple ASR→LLM→TTS loop to produce spoken replies, typically in a few seconds. The Rust server is the technical heart: configurable via config.toml to route audio to ASR (e.g., Whisper), send transcripts to an LLM, and pass the output to a TTS model. The stack supports modular chains (ASR/LLM/TTS), end-to-end models like Gemini, and an MCP extension for custom actions or home automation. This makes EchoKit notable for the AI/ML community because it packages a reproducible, low-cost, privacy-friendly voice agent platform that’s easy to customize, extend, and run locally — ideal for makers, classrooms, and developers experimenting with model selection, latency, and embedded voice workflows.