query_codebase
AI agents call query_codebase to retrieve information from Codesearch without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
The tool retrieves and searches code from a monorepo without modifying, executing, or deleting anything. Sibling tools like 'query_single_file' and 'service_status' further confirm this is a query-oriented server. Even though the tool description is empty, the server's explicit purpose as a code search interface and the naming convention clearly indicate a Read operation.
From the tool's definition Tool name 'query_codebase' and server description indicate 'Full-text and structural code search' with results 'exposed as MCP tools so Claude can query the codebase directly'. This is a read-only search/query operation with no side effects.
Attacks that exploit this kind of access
query_codebase. It is categorised as a Read tool in the Codesearch MCP Server, which means it retrieves data without modifying state.
Register the Codesearch MCP server in PolicyLayer and add a rule for query_codebase: 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 Codesearch. Nothing to install.
query_codebase 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 query_codebase 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 query_codebase. 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.
query_codebase is provided by the Codesearch MCP server (microsoft/tscodesearch). 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 →