Loading ToolBench
MCP server combining Graphiti knowledge graph with Mem0 memory layer for long-term knowledge storage and short-term memory management
Server has 10 tools with complete schemas and descriptions. Naming follows verb_noun convention (add_episode, search_facts, get_entities, etc.). Descriptions are present and contextual (avg ~120 chars), explaining WHAT and WHEN. However, parameter descriptions lack detail on constraints, formats, and valid ranges. Output schemas are not documented, LLMs cannot predict response structure. Error handling is absent from tool definitions. No tool annotations (readOnlyHint/destructiveHint) despite clear risk levels. Destructive tools (delete_memory, delete_all_memories) lack confirmation patterns or dry-run support.
Add an episode (text content) to the knowledge graph. Use this to store facts, conversations, or any knowledge that should be remembered long-term with relationships.
Add a memory from conversation messages. Use this for user preferences, recent context, and information that should be recalled quickly with minimal tokens.
Delete all memories for a user. Use with caution.
Delete a specific memory by ID.
Get all entity nodes from the knowledge graph. Entities are people, places, concepts, or things that have been extracted.
Get all memories for a user.
No output schemas documented. LLMs cannot predict response structure (fields, types, pagination). Forces agents to guess what data is returned and how to chain tools.
Destructive tools (delete_memory, delete_all_memories) lack confirmation patterns, dry-run support, or explicit warnings in descriptions. Agents can irreversibly delete data without safeguards.
Parameter descriptions lack constraint details. 'limit' params have no min/max bounds (e.g., can agent pass limit=999999?). 'content' in add_episode has no length guidance. 'messages' array in add_memory has no item count limits.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 53 | <=2025-11-25 | v2 |
Get relationships for a specific entity from the knowledge graph.
Search for facts and relationships in the knowledge graph. Returns entities and their relationships based on semantic and keyword matching.
Search for relevant memories. Returns the most relevant stored memories based on semantic similarity. Token-efficient retrieval.
Update an existing memory by ID.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Agents cannot distinguish safe reads from destructive writes without parsing descriptions. Violates current MCP spec (2026-07-28).
No error handling guidance in tool definitions. No recovery hints (e.g., 'If entity not found, try search_facts() first'). Agents have no actionable next steps on failure.