AI agents call lookup-subjects to retrieve information from Spicedb without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool retrieves permission information from SpiceDB without modifying, deleting, or executing external operations. It fits the 'Read' category as it queries and returns data about which subjects have permissions on resources. The blast radius of misuse is limited to information disclosure about the permission model.
From the tool's definition Tool 'lookup-subjects' queries and retrieves data (finds all subjects with a permission on a resource). The description explicitly states it is used for 'answering questions', which is a read operation against SpiceDB.
Attacks that exploit this kind of access
Finds all subjects of a specified type that have a particular permission on a resource. This is useful for answering questions like. It is categorised as a Read tool in the Spicedb MCP Server, which means it retrieves data without modifying state.
Register the Spicedb MCP server in PolicyLayer and add a rule for lookup-subjects: 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 Spicedb. Nothing to install.
lookup-subjects 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 lookup-subjects 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 lookup-subjects. 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.
lookup-subjects is provided by the Spicedb MCP server (samkim/spicedb-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 →