Semantic search with composite relevance scoring.
AI agents call memory_search to retrieve information from Claude Memory MCP without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool retrieves or queries data (performs semantic search across stored memory) with no side effects, state changes, or reversible modifications. It matches the Read category definition of 'retrieves or queries data; no side effects (search, list, get, fetch).' The search-based operation has minimal blast radius if misused—at worst, an agent retrieves unintended memory content, but no data is created, modified,…
From the tool's definition Tool name 'memory_search' and description 'Semantic search with composite relevance scoring' indicate retrieval of data without modification or deletion. The verb 'search' is a classic read operation that queries existing memory entries.
Attacks that exploit this kind of access
Semantic search with composite relevance scoring. It is categorised as a Read tool in the Claude Memory MCP MCP Server, which means it retrieves data without modifying state.
Register the Claude Memory MCP server in PolicyLayer and add a rule for memory_search: 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 Claude Memory MCP. Nothing to install.
memory_search 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 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. 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 is provided by the Claude Memory MCP server (navid-kianfar/claude-memory-mcp). 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 →