AI agents call search_code to retrieve information from Brain without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool retrieves stored semantic memory and knowledge linked to code through filtering and matching. It has no side effects, does not modify data, execute code, or trigger external operations. It is purely a query/search function that reads from the PostgreSQL backend. The blast radius of misuse is minimal—an agent could retrieve irrelevant or sensitive stored knowledge, but cannot alter or act on it.
From the tool's definition Tool description states it 'Search[es] knowledge' and 'uses semantic matching' - these are query/retrieval operations. No modification, deletion, or execution capabilities are mentioned.
Attacks that exploit this kind of access
Search knowledge linked to code. Filter by repository, file path, or symbol name. Uses semantic matching on the knowledge content. It is categorised as a Read tool in the Brain MCP Server, which means it retrieves data without modifying state.
Register the Brain MCP server in PolicyLayer and add a rule for search_code: 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 Brain. Nothing to install.
search_code 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_code 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_code. 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_code is provided by the Brain MCP server (markschaake/brain-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 →