MCP server for IoT device control via MQTT with long-term memory storage and retrieval using Mem0
Server has 7 tools with significant quality gaps. Tool naming is inconsistent (turn_on_light vs turn_off_light vs toggle_light vs list_devices). Descriptions exist but are verbose and lack actionable guidance. Critical issue: 3 tools (turn_on_light, turn_off_light, toggle_light) have a dummy parameter with description 'Optional parameter that does nothing (required for schema compatibility)', this is a red flag indicating schema design problems. Input schemas are present but poorly designed. No output schemas documented. Error handling is minimal. Memory tools (save_memory, search_memories) lack parameter constraints and validation guidance.
Get all stored memories for the user. Call this tool when you need complete context of all previously memories. Returns a JSON formatted list of all stored memories, including when they were created and their content. Results are paginated with a default of 50 items per page.
List all connected IoT devices. This tool returns information about all IoT devices that have been discovered on the MQTT network. For each device, it provides the ID, type, online status, and available commands. Note: Do NOT try to call a device ID directly. Instead, use the command tools like: - light_bulb_on - To turn the light bulb ON - light_bulb_off - To turn the light bulb OFF - light_bulb_toggle - To toggle the light bulb - check_light_status - To check light status
Save information to your long-term memory. This tool is designed to store any type of information that might be useful in the future. The content will be processed and indexed for later retrieval through semantic search.
Search memories using semantic search. This tool should be called to find relevant information from your memory. Results are ranked by relevance. Always search your memories before making decisions to ensure you leverage your existing knowledge.
Dummy parameter anti-pattern: turn_on_light, turn_off_light, toggle_light all require a 'dummy' string parameter with description 'Optional parameter that does nothing (required for schema compatibility)'. This violates single-responsibility and confuses LLMs about what the tool actually needs.
No output schemas documented. Tools return free-text strings (e.g., list_devices returns markdown-like text). LLMs cannot parse structured results or chain outputs to downstream tools. Violates response-shaper pattern.
Memory tools lack input validation constraints. save_memory accepts any string with no length limits or content guidelines. search_memories accepts 'limit' as integer with no min/max bounds (could be 0 or 1000000). Violates constrained-input pattern.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 48 | 2026-07-28+ | v2 |
Toggle the light bulb on/off. This is a simplified tool to toggle the light bulb power state.
Turn OFF the light bulb. This tool will explicitly turn off the light bulb, setting its power state to false.
Turn ON the light bulb. This tool will explicitly turn on the light bulb, setting its power state to true.
Descriptions are verbose and lack actionable guidance. list_devices description is 280+ chars and includes redundant notes about not calling device IDs directly. Descriptions should be 10-200 chars with clear WHAT/WHEN/WHY structure.
No error handling or recovery guidance. Tools return generic strings on failure (e.g., 'Found light_bulb_001 device. Use light_bulb_on or light_bulb_off to control it.'). No indication of what went wrong, whether to retry, or what to do next.