Back to verdicts
MCP serverPublished

MCP Memory Reference Server

The Memory reference server passed a real MCP smoke test for entity creation and search. It is useful as a simple knowledge-graph memory primitive, but the test did not validate persistence guarantees, conflict behavior, or controls for sensitive memory use.

Tested 2026-06-11sc-agent-trust-v0.1Subject page

Independent trust badge

The visible trust mark for this verdict.

SilentCritique verdict badge for MCP Memory Reference Server

Badge clicks resolve to this canonical verdict so the score, test date, evidence, limitations, and reply status remain attached.

Embed

Show this badge on your site

[![SilentCritique verdict for MCP Memory Reference Server](https://silentcritique.com/badges/mcp-memory-reference)](https://silentcritique.com/verdicts/mcp-memory-reference)

Markdown works in GitHub READMEs. The badge always links back to this verdict.

Editorial notice

This page reflects SilentCritique's independent editorial opinion based on the specific test evidence shown. It is not an allegation of unlawful, malicious, fraudulent, or bad-faith conduct. SilentCritique does not accept payment to remove criticism, change a score, suppress a verdict, or improve an outcome.

Claim tested

Can the public Memory MCP server create a knowledge-graph entity and retrieve it through search in a real MCP session?

Evaluator panel

Protocol harnessData-safety reviewerOperator skeptic

Evidence reviewed

Nine memory tools discovered

The server exposed graph read, search, create, relation, observation, and delete tools.

evidence/trust5/2026-06-11-mcp-pilot.json

Entity creation succeeded

The harness created a SilentCritique Trust 5 evaluation_run entity through create_entities.

evidence/trust5/2026-06-11-mcp-pilot.json

Search retrieved the created entity

A search_nodes query for SilentCritique Trust 5 returned the entity and observation created earlier in the same run.

evidence/trust5/2026-06-11-mcp-pilot.json

Test setup

  • Started @modelcontextprotocol/server-memory@2026.1.26 over MCP stdio.
  • Used the official MCP client SDK to list tools, create an evaluation_run entity, and search for it.
  • Stored the full tool-call evidence in evidence/trust5/2026-06-11-mcp-pilot.json.

Strengths

  • Basic graph write and retrieval worked in one MCP session.
  • The tool surface is compact and easy for clients to inspect.
  • Structured content was returned alongside text output for the tested calls.

Failure modes

  • Memory tools can store sensitive user data if a host routes private context into them without policy.
  • Delete and mutation tools need client-side governance in shared environments.
  • The test did not establish durability, encryption, tenancy isolation, or conflict resolution.

What would improve the score

  • Document safe defaults for sensitive memory handling.
  • Add explicit retention and deletion guidance for host applications.
  • Expose a repeatable persistence test in the reference documentation.

Limitations

  • This was an unsolicited smoke test of the public package, not a privacy or storage audit.
  • Only entity creation and search were tested.
  • No restart, multi-client, or large-graph behavior was evaluated.

Visible dissent

  • The protocol harness scored the server well because the core create/search loop worked.
  • The data-safety reviewer withheld points because memory systems carry privacy risk beyond basic tool success.

Right of reply

No vendor reply has been requested or published as of 2026-06-11. SilentCritique will publish factual corrections or a right of reply through the corrections process.

Methodology matters

Scores are only meaningful when the rubric, date, evidence, and dissent are visible.

Read methodology