AI agents call search_book to retrieve information from Narrarium without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool retrieves and queries data from a local repository without side effects. It enables searching/filtering existing book content but does not create, modify, delete, or execute operations. Classification as Read is appropriate with low severity since information retrieval from a local repository poses minimal risk even if an AI agent misuses it.
From the tool's definition Tool name 'search_book' and description indicate it 'Search[es] the local book repository across canon, chapters, notes, summaries, and research' with no mention of modification, deletion, or execution capabilities.
Attacks that exploit this kind of access
Search the local book repository across canon, chapters, notes, summaries, and research before drafting or editing. It is categorised as a Read tool in the Narrarium MCP Server, which means it retrieves data without modifying state.
Register the Narrarium MCP server in PolicyLayer and add a rule for search_book: 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 Narrarium. Nothing to install.
search_book 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_book 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_book. 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_book is provided by the Narrarium MCP server (narrarium-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 →