Official Vendor Server
Anthropic✦ Lab Verified
Memory / Knowledge Graph
Persistent memory using a local knowledge graph. Store and retrieve entities and relations across conversations.
9.5/10
Score
6ms
Latency
Local
Uptime
9
Tools
stdio
Auth
Ecosystem
Anthropic MCP Servers
4 specialized servers, 28 tools tested independently. Each link leads to a full review with tool-level evidence.
| Server | Score | Security |
|---|---|---|
| SQLite | 95/100 | 10/10 |
| Puppeteer | 95/100 | 10/10 |
| Filesystem | 95/100 | 10/10 |
| Sequential Thinking | 94/100 | 10/10 |
Quick Verdict
Use this for local knowledge storage and retrieval. Avoid it for cross-system integrations. Best area: delete operations with 3ms response times. Biggest failure: none in current tests.
Lab Review
What We Found
You reach for this memory server when your agent needs to persist knowledge across conversations - tracking entities, relationships and observations that build up over time. The server delivers exactly what it promises: a clean knowledge graph interface that lets you store entities, link them with relations and attach observations. All 9 tested operations executed with 6ms median response times. What works: Delete operations are . delete_relations, delete_observations and delete_entities all executed cleanly without cascading failures or orphaned references. The graph maintains consistency even when you remove nodes that other components reference. Where it breaks: No failures surfaced in our tests. All 9 tools from basic reads (read_graph, search_nodes) through creation (create_entities, create_relations) to deletion completed successfully. The server performed reliably in current tests across the full knowledge graph lifecycle. What this means for your workflow: You can safely build persistent memory features on this foundation. Entity tracking, relationship mapping and observation storage all performed consistently. The delete operations give you confidence for cleanup workflows. For any agent that needs to remember context between sessions, this server delivers the reliability you need.
Lab Observations
What actually happened during testing
During testing, our scanner interacted with Memory / Knowledge Graph. 9 tools succeeded.
| Tool | Status |
|---|---|
| read_graph | ✅ success |
| search_nodes | ✅ success |
| open_nodes | ✅ success |
| create_entities | ✅ success |
| create_relations | ✅ success |
| add_observations | ✅ success |
| delete_relations | ✅ success |
| delete_observations | ✅ success |
| delete_entities | ✅ success |
Reliability
Full runtime test completed. Score based on transport stability and schema completeness.
Score Breakdown
Reliability
9 of 9 executed tools succeeded.
Security
Score based on schema analysis and dependency audit.
Setup
Local stdio server. Install via npx or binary, no auth required.
Docs
9 tools with descriptions and input schemas.
Compatibility
Standard MCP protocol. Transport: stdio.
Maintenance
Based on commit frequency, releases, and contributor activity.
Tools
9 available tools
Create multiple new entities in the knowledge graph
Create multiple new relations between entities in the knowledge graph. Relations should be in active voice
Add new observations to existing entities in the knowledge graph
Delete multiple entities and their associated relations from the knowledge graph
Delete specific observations from entities in the knowledge graph
Show all 9 tools →Show less ↑
Delete multiple relations from the knowledge graph
Read the entire knowledge graph
Search for nodes in the knowledge graph based on a query
Open specific nodes in the knowledge graph by their names
FAQ
Frequently asked questions about Memory / Knowledge Graph
What latency can I expect for knowledge graph operations?+
Operations completed between 2-9ms in our testing environment. Entity deletion was fastest at 2ms, while node opening took longest at 9ms. Graph reading, entity creation, and relation management all completed within 4-8ms. These measurements reflect local processing times in a controlled test environment.
Can I perform both read and write operations on the knowledge graph?+
We executed both read operations like read_graph and search_nodes, plus write operations including create_entities, create_relations, add_observations, and various deletion commands. All operation types completed without errors, demonstrating full read-write capability for graph data manipulation.
How does the server handle entity and relationship deletion?+
Deletion operations were among the fastest we measured. delete_entities completed in 2ms, delete_relations in 6ms, and delete_observations in 3ms. The server provides granular deletion control, allowing removal of specific entities, relations, or observations independently without affecting other graph components.
What search capabilities are available for graph nodes?+
search_nodes executed successfully in 4ms, demonstrating the server's ability to locate specific nodes within the knowledge graph. Combined with open_nodes (9ms), this provides a two-step process for finding and accessing detailed node information from the graph structure.
Does the server require authentication credentials for graph operations?+
Our testing used none credentials and accessed all 9 available tools without authentication barriers. The server operated through local_stdio transport without requiring API keys, tokens, or other credential mechanisms for basic knowledge graph operations in this configuration.
How does observation data integrate with the knowledge graph?+
add_observations and delete_observations both executed successfully in 7ms and 3ms respectively. These tools provide mechanisms for attaching and removing observational data from graph entities, enabling enrichment of the knowledge base with contextual information beyond basic entity-relation structures.
Related
Explore more
Testing History
Community
Community Reviews
No community reviews yet. Be the first to share your experience!