Mijia

**Quick judgment** A focused and fairly capable skill for controlling Xiaomi Mijia lamps (and potentially other Mijia devices) through mijiaAPI. It is best suited for single-lamp or small desktop setups where detailed control (brightness, color temperature, and modes) is desired. **Strengths** - Good coverage of lamp capabilities: on/off, brightness, color temperature, and multiple preset modes. - Well-defined CLI interface and command mapping that an LLM agent can use deterministically. - Natural language triggers for common actions (turn on/off, brighter/dimmer, warm/cool light, reading/computer/night modes). **Limitations & risks** - Primarily documented for a single lamp via `MIJIA_LAMP_DID`; multi-device or non-lamp control is not clearly specified. - Depends on Xiaomi account login and mijiaAPI; reliability is tied to Xiaomi’s cloud and token/session stability. - No explicit safeguards against unintended changes (e.g., misinterpreting vague user requests could turn lights on/off unexpectedly). - Setup requires environment variables and a QR-based login flow, which may be error-prone for less technical users. **Recommended scenarios** - Controlling a Xiaomi Mijia desk lamp in a home office or study area via voice/chat. - Automating routine lighting adjustments (switch to reading or computer mode, night mode before bed, etc.). - Context-aware lighting tweaks where the agent first checks lamp status and then adjusts brightness or temperature. - Use in environments where the operator is comfortable with initial technical setup (env vars, QR login, `uv`/Python tooling).