AI agents call search_documents to retrieve information from Local Rag 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 knowledge base to answer questions. It generates responses based on semantic search of indexed documents but does not create, modify, delete, or execute external operations. This is a pure read operation with no side effects on the document store or external systems.
From the tool's definition Tool description indicates 'RAG 検索を行い' (performs RAG search) and '回答を生成します' (generates answers). The sibling tools are add_document, list_documents, and this tool searches/queries documents.
Attacks that exploit this kind of access
質問に対して RAG 検索を行い、qwen3.5:9b で回答を生成します。. It is categorised as a Read tool in the Local Rag MCP Server, which means it retrieves data without modifying state.
Register the Local Rag MCP server in PolicyLayer and add a rule for search_documents: 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 Local Rag. Nothing to install.
search_documents 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_documents 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_documents. 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_documents is provided by the Local Rag MCP server (n-irei/local-rag-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 →