MCP server that exposes memory search tool for Claude to find relevant context from past sessions stored in EverOS/EverMind Cloud
Single tool 'evermem_search' has a reasonable description (95 chars) and basic schema structure, but critical gaps prevent higher scoring. The schema is present with proper types and required fields, but lacks depth in parameter descriptions. The tool description mentions use cases but is generic. No error handling guidance, no output schema documentation, and no tool annotations. The server exposes only a read-only search interface with minimal composition potential. Baseline: 549 production tools average 194 chars in descriptions and 100% have documented return types, this server falls short on both counts.
Search past conversation memories. Returns summaries with dates and relevance scores. Use when user asks about previous work, decisions, or context from past sessions. Params: query (required), limit (default: 10, max: 20)
Output schema not documented. Tool description does not specify what fields are returned (e.g., memory summaries, dates, relevance scores structure). LLMs cannot plan downstream operations without knowing response shape.
Parameter descriptions are minimal. 'query' lacks format guidance (free-form keywords? natural language questions?). 'limit' description is bare, no mention of performance implications or when to use smaller limits.
No error handling guidance. Tool description does not explain what happens on empty results, invalid queries, or service failures. LLMs have no recovery path.
Tool name 'evermem_search' is domain-specific jargon. Clearer names like 'search_memories' or 'search_conversation_history' would better signal intent to LLMs unfamiliar with EverMem terminology.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 57 | 2026-07-28+ | v2 |
No pagination or result limiting enforcement in schema. Tool accepts 'limit' but schema does not enforce max=20 constraint at the JSON Schema level, only in description text, which LLMs may ignore.