AI agents call kafka_get_watermarks to retrieve information from MCP Kafka without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool retrieves information about Kafka topic partition watermarks and message counts. It performs a query operation with no side effects, no data modification, and no external command execution. The blast radius of misuse is minimal—an agent could only view partition metadata, not alter topics, messages, or cluster state.
From the tool's definition Tool name 'kafka_get_watermarks' and description 'Get low/high watermarks and message counts for all partitions of a topic' indicate a read-only query operation that retrieves metadata about Kafka partitions without modifying any state.
Attacks that exploit this kind of access
Get low/high watermarks and message counts for all partitions of a topic. It is categorised as a Read tool in the MCP Kafka MCP Server, which means it retrieves data without modifying state.
Register the MCP Kafka MCP server in PolicyLayer and add a rule for kafka_get_watermarks: 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 Kafka. Nothing to install.
kafka_get_watermarks 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 kafka_get_watermarks 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 kafka_get_watermarks. 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.
kafka_get_watermarks is provided by the MCP Kafka MCP server (williajm/mcp_kafka). 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 →