Get all unread messages (direct and from subscribed channels). Marks them as read. Messages with your name in the
AI agents call check_messages to retrieve information from Agent Comms without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool retrieves and queries existing messages without creating, modifying, deleting, or executing any external operations. The marking as read is a metadata update that does not alter message content or cause destructive effects. It poses minimal risk when misused by an AI agent, as it only exposes already-accessible communication data.
From the tool's definition Tool name is 'check_messages' and description states 'Get all unread messages (direct and from subscribed channels).
Attacks that exploit this kind of access
Get all unread messages (direct and from subscribed channels). Marks them as read. Messages with your name in the. It is categorised as a Read tool in the Agent Comms MCP Server, which means it retrieves data without modifying state.
Register the Agent Comms MCP server in PolicyLayer and add a rule for check_messages: 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 Agent Comms. Nothing to install.
check_messages 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 check_messages 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 check_messages. 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.
check_messages is provided by the Agent Comms MCP server (mavericky007/agent-comms-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 →