MCP server for Alexa home automation control, providing tools to manage smart home devices including lights, volume, announcements, sensors, and Do Not Disturb settings
This server has 14 tools with mixed definition quality. Naming is generally strong and verb-driven (get_, set_, list_, adjust_). Descriptions are present and contextual, though several are generic or missing key details about behavior and return values. Input schemas are visible and mostly typed, but output schemas are undocumented across all tools. Several parameters lack full constraint definitions (e.g., no min/max for numeric ranges, enums for device selection). Error handling is minimal, no recovery guidance, retryability classification, or self-correction hints. Security posture appears sound (no visible credentials in params), but audit trails and permission gating are not evident. Tool composition is reasonable but some tools could be better integrated (e.g., set_light_* tools require separate calls when a batch variant would be more efficient).
Adjust volume level by amount (-100 to +100) for Alexa device. If you have only one device, no deviceType/dsn needed - it will be auto-detected.
Send voice announcements to Alexa devices with smart suppression
Get comprehensive sensor readings including temperature, illuminance, motion detection, and light status from all available sensors
Get current bedroom temperature, illuminance, and light status for context-aware decisions
Get current volume levels for all Alexa devices
Get Do Not Disturb status for all Alexa devices
Output schemas completely undocumented. No tool specifies what fields are returned, their types, or structure. LLMs cannot determine how to extract or chain data from responses.
Numeric parameters lack explicit range/constraint documentation. Brightness, volume, and adjustment amounts have no stated min/max in descriptions (e.g., 'brightness 0-100' is in the description but not enforced or formally constrained). LLMs may pass invalid values; no validation error guidance.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-21 | D | 58 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 13 | - | v1 |
Get current music playbook status from Alexa/Spotify integration
Get all available smart lights and their capabilities. Shows how many lights you have and their IDs for reference.
List all available smart home devices with their capabilities, categories, and status information
Set volume level (0-100) for Alexa device. If you have only one device, no deviceType/dsn needed - it will be auto-detected.
Enable or disable Do Not Disturb mode for an Alexa device. If no device specified, uses the primary Echo device.
Set smart light brightness (0-100%). If you have only one light, no ID needed - it will be auto-detected.
Set smart light color by name or Kelvin temperature. Supports full color spectrum and white shades. If you have only one light, no ID needed.
Turn smart light on or off. If you have only one light, no ID needed - it will be auto-detected.
Optional parameters (e.g., 'id' in set_light_power, deviceType/dsn in set_device_volume) are marked optional but descriptions state 'auto-detected if only one'. This behavior is underdocumented, what happens if user has multiple devices and omits the parameter? Error or silent fallback? No clear recovery guidance.
No error handling or recovery guidance documented. How does the tool respond to invalid device IDs? Missing sensors? Connection failures? LLMs have no guidance on retry logic, fallbacks, or what to suggest to users.
Destructive/write operations (set_light_*, set_device_volume, adjust_device_volume, set_dnd_status, alexa_announce) lack confirmation or dry-run modes. Agents may execute these inadvertently, causing unintended state changes (e.g., turning off all lights, broadcasting announcements).
Discovery tools (list_lights, list_smarthome_devices, get_all_sensor_data, get_device_volumes) return no documented result limits or pagination. Large device sets may exceed context windows; descriptions don't guide when to call these upfront vs. on-demand.
Device selection across tools is inconsistent. Some use 'id', others 'deviceType'/'dsn', others 'deviceSerialNumber'. No unified identifier convention. LLMs must reason separately for each tool rather than following a predictable pattern.