Semantic search across an Anima
AI agents call search_memories to retrieve information from Elephantasm without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool performs a query/search operation on persistent memory data. It retrieves information based on semantic similarity without creating, modifying, deleting, or executing code. The blast radius of misuse is minimal—an agent could search for sensitive information but cannot alter or destroy it. Classified as Read with low severity.
From the tool's definition Tool name 'search_memories' and description indicate semantic search functionality across stored memory. The word 'search' combined with the context of a memory/knowledge system indicates data retrieval with no modification capability.
Attacks that exploit this kind of access
Semantic search across an Anima. It is categorised as a Read tool in the Elephantasm MCP Server, which means it retrieves data without modifying state.
Register the Elephantasm MCP server in PolicyLayer and add a rule for search_memories: 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 Elephantasm. Nothing to install.
search_memories 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 search_memories 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 search_memories. 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.
search_memories is provided by the Elephantasm MCP server (kaminocorp/elephantasm-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 →