Execute a single read-only SELECT against whitelisted views (v_active_subscriptions, v_monthly_revenue, v_customer_summary). Rejects writes, JOINs, and base-table access. Returns the exact SQL executed, with PII masked and rows capped at 500.
AI agents call safe_query to retrieve information from Semantic BI MCP without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool is purely a data retrieval mechanism with multiple safety constraints: read-only semantics, whitelist enforcement, write rejection, row capping, and PII masking. It queries predefined views rather than allowing arbitrary SQL construction. The blast radius of misuse is limited to unauthorized data access (not modification or deletion), making it a Read category tool with low severity.
From the tool's definition Tool name 'safe_query' and description explicitly states 'read-only SELECT' restricted to 'whitelisted views', rejects 'writes', enforces row limits (500), and masks PII. No modification, deletion, or external execution capability.
Attacks that exploit this kind of access
Execute a single read-only SELECT against whitelisted views (v_active_subscriptions, v_monthly_revenue, v_customer_summary). Rejects writes, JOINs, and base-table access. Returns the exact SQL executed, with PII masked and rows capped at 500. It is categorised as a Read tool in the Semantic BI MCP MCP Server, which means it retrieves data without modifying state.
Register the Semantic BI MCP server in PolicyLayer and add a rule for safe_query: 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 Semantic BI MCP. Nothing to install.
safe_query 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 safe_query 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 safe_query. 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.
safe_query is provided by the Semantic BI MCP server (lovrobubanic/semantic-bi-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 →