AI agents call recall to retrieve information from Claude without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool performs semantic search over stored memory to retrieve contextually relevant information. It is a pure read operation that queries and returns data without side effects, modifications, or execution of code. The ability to retrieve context across sessions poses minimal risk even if misused by an agent, as it only accesses stored information rather than modifying, deleting, or executing operations.
From the tool's definition Tool name 'recall' and description 'Recupera fragmentos relevantes de memoria por similitud semántica' (retrieves relevant memory fragments by semantic similarity) indicate a retrieval operation with no modification or execution of external operations.
Attacks that exploit this kind of access
Recupera fragmentos relevantes de memoria por similitud semántica. It is categorised as a Read tool in the Claude MCP Server, which means it retrieves data without modifying state.
Register the Claude MCP server in PolicyLayer and add a rule for recall: 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. Nothing to install.
recall 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 recall 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 recall. 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.
recall is provided by the Claude MCP server (zcc-cddg-mx/mcp-server-claude). 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 →