Give your AI a memory. MCP server for persistent context across all sessions.
Memory Buddy defines 3 tools with basic schemas and descriptions, but falls short of production quality in multiple dimensions. Tool names use verb_noun pattern correctly (memory_status, memory_search, memory_compact), but descriptions are minimal (34 - 53 chars, below the 50 - 200 char LLM-optimized baseline). Input schemas are present but lack parameter descriptions for critical fields. Output schemas are not documented. No error handling guidance or recovery paths. No tool annotations (readOnlyHint/destructiveHint). Missing pagination for search results. Parameter descriptions are sparse or absent.
Force compaction of current session
Search through memories by keyword
Get memory system status (events count, last session, etc.)
Descriptions are too short (30 - 53 chars). LLM-optimized descriptions should be 50 - 200 chars and answer WHAT, WHEN, and WHY. Current descriptions lack context for tool selection and downstream behavior.
memory_search lacks pagination support (page/offset, limit, total_count). Unbounded search results can exhaust context windows. No limit stated in description.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). memory_status and memory_search should declare readOnlyHint: true; memory_compact should declare destructiveHint: true. Agents cannot infer side-effect risk.
Output schemas not documented. Callers do not know what fields memory_status, memory_search, or memory_compact return. Agents cannot plan downstream tool calls or extract relevant data.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 40 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 0 | - | v1 |
No error handling guidance. If memory_search finds nothing, or memory_compact fails, responses do not tell the LLM what to do next. No recovery paths (e.g., 'Try a different query', 'Check storage permissions').
memory_search input parameter 'query' has minimal description ('Search query'). Does it support keywords, phrases, boolean operators? Does it return relevance scores? Case-sensitive? Fuzzy matching? No guidance.
memory_status and memory_compact accept zero parameters but no description hints at what stateful context they operate on. Session ID or memory store path should be documented or parameterizable.