AI agents call semantic_search to retrieve information from Rewind 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 personal reading history data based on semantic similarity. It has no side effects, does not create, modify, or delete data, and does not execute code or trigger external operations. It fits the 'Read' category as a search/query operation.
From the tool's definition Tool name is 'semantic_search' and description explicitly states it performs 'Semantic search over the reading domain' — a query operation with no modification of data.
Attacks that exploit this kind of access
Semantic search over the reading domain using Voyage AI embeddings. Use when the user describes the gist or topic of an article they remember rather than quoting exact words — e.g. It is categorised as a Read tool in the Rewind MCP Server, which means it retrieves data without modifying state.
Register the Rewind MCP server in PolicyLayer and add a rule for semantic_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 Rewind. Nothing to install.
semantic_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 semantic_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 semantic_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.
semantic_search is provided by the Rewind MCP server (rewind-mcp-server). 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 →