执行只读 SQL 查询（例如 SELECT/SHOW/PRAGMA）。
AI agents call read_sql to retrieve information from MCP Database Manager 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 data from databases without modifying, deleting, or executing code. The explicit mention of only SELECT/SHOW/PRAGMA statements confirms it has no capability to create, modify, or delete data. The severity is low because read operations have minimal blast radius—the worst outcome is information disclosure of data the agent is already authorized to query.
From the tool's definition Tool name is 'read_sql' and description states '执行只读 SQL 查询（例如 SELECT/SHOW/PRAGMA）' (Execute read-only SQL queries such as SELECT/SHOW/PRAGMA). These are canonical read-only operations with no side effects.
Attacks that exploit this kind of access
执行只读 SQL 查询（例如 SELECT/SHOW/PRAGMA）。. It is categorised as a Read tool in the MCP Database Manager MCP Server, which means it retrieves data without modifying state.
Register the MCP Database Manager MCP server in PolicyLayer and add a rule for read_sql: 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 MCP Database Manager. Nothing to install.
read_sql 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 read_sql 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 read_sql. 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.
read_sql is provided by the MCP Database Manager MCP server (meimingqi222/mcp-database-manager). 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 →