AI agents call memory_search_semantic to retrieve information from Mementos without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool queries existing memory data using vector embeddings or TF-IDF to find conceptually similar items. It retrieves information without side effects, making it a Read operation. The low severity reflects that misuse would only expose stored memories rather than modify, delete, or execute arbitrary actions. The high confidence is justified by the explicit description emphasizing search and retrieval semantics.
From the tool's definition Tool performs 'Semantic (meaning-based) memory search' and 'Finds memories by conceptual similarity' — purely retrieval operations with no data modification, deletion, or code execution.
Attacks that exploit this kind of access
Semantic (meaning-based) memory search using vector embeddings. Finds memories by conceptual similarity, not keyword match. Uses OpenAI embeddings if OPENAI_API_KEY is set, otherwise TF-IDF. It is categorised as a Read tool in the Mementos MCP Server, which means it retrieves data without modifying state.
Register the Mementos MCP server in PolicyLayer and add a rule for memory_search_semantic: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Mementos. Nothing to install.
memory_search_semantic is a Read tool with low risk. Read-only tools are generally safe to allow by default.
Yes. Add a rate_limit block to the memory_search_semantic rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for memory_search_semantic. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
memory_search_semantic is provided by the Mementos MCP server (@hasna/mementos). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Every MCP server has a record like this.
Type a name, get the same breakdown: verified identity, auth posture, risk grade, capabilities, recommended policy.
Teams ship this data inside their own products. See what a licence covers →