What problem does it solve? Developing device-side firmware for the ESP Private Agents AI platform requires exact knowledge of the esp-agents-firmware SDK's initialization order, event model, tool registration contract, and board configuration; guessing APIs leads to broken builds and non-starting agents. This Skill grounds every generated line in the repository's real headers, Kconfig symbols, and example source so firmware compiles and connects correctly. ## Core Features & Use Cases - Scenario Recipes: Twelve step-by-step recipes covering build/flash, custom boards, custom deployments, agent initialization and events, provisioning, local tool development, built-in tools, Matter controller service and control, audio pipeline, console commands, and headless devices. - Verified API & Config Reference: Real signatures for esp_agent, app_agent, agent_setup, app_audio, app_device, app_display, and Matter controller modules, plus all Kconfig options and board-level macros. - Pitfall Prevention: Thirteen critical pitfalls with wrong/correct code pairs, such as heap-allocating tool results, forwarding events to the default handler, and satisfying the three agent-start preconditions. - Use Case: Add a new local tool to a voice chat agent on an ESP-BOX-3: register the handler in firmware, declare the matching inputSchema in agent_config.json, follow the canonical app_main init order, then build and flash with idf.py. ## Quick Start Ask the AI to create a voice chat agent firmware based on the voice_chat example for the ESP-BOX-3 board, including a custom local tool registered in both firmware and agent_config.json.
