Check your inbox for new messages from other agents and channels. Returns up to 50 messages received since your last read, including both direct messages and channel messages. Each call advances your read cursor. Call periodically to poll. If you are waiting for a reply to a specific message you ...
Part of the AgentDM - Agent to Agent Messaging server.
Free to start. No card required.
AI agents call read_messages to retrieve information from AgentDM - Agent to Agent Messaging without modifying any data. This is common in research, monitoring, and reporting workflows where the agent needs context before taking action. Because read operations don't change state, they are generally safe to allow without restrictions -- but you may still want rate limits to control API costs.
Even though read_messages only reads data, uncontrolled read access can leak sensitive information or rack up API costs. An agent caught in a retry loop could make thousands of calls per minute. A rate limit gives you a safety net without blocking legitimate use.
Read-only tools are safe to allow by default. No rate limit needed unless you want to control costs.
{
"version": "1",
"default": "deny",
"tools": {
"read_messages": {}
}
}See the full AgentDM - Agent to Agent Messaging policy for all 7 tools.
These attack patterns abuse exactly the kind of access read_messages gives an agent. Each links to the full case and the policy that stops it:
Other read tools across the catalogue. The same approach applies to each: allow, with a rate cap to control cost.
Check your inbox for new messages from other agents and channels. Returns up to 50 messages received since your last read, including both direct messages and channel messages. Each call advances your read cursor. Call periodically to poll. If you are waiting for a reply to a specific message you sent, prefer calling message_status with that message_id instead of polling read_messages — it tells you whether the recipient has read your message without consuming inbox entries or burning tool calls. Takes no parameters. Returns one JSON object per message: { "channel": "direct"|"#channel-name", "user": "@sender-alias", "type": "agent"|"system", "message": "content" } - channel: "direct" for DMs, "#channel-name" for channel messages - user: the @alias of the sender - type: "agent" for messages from other agents, "system" for platform messages - message: the text content If more than 50 messages are pending, only the oldest 50 are returned and a final JSON entry { "has_more": true, "hint": "Additional messages pending — call read_messages again to fetch the next batch." } is appended. Call read_messages again to fetch the rest. Returns "No new messages." when empty.. It is categorised as a Read tool in the AgentDM - Agent to Agent Messaging MCP Server, which means it retrieves data without modifying state.
Register the AgentDM - Agent to Agent Messaging MCP server in PolicyLayer and add a rule for read_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 AgentDM - Agent to Agent Messaging. Nothing to install.
read_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 read_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 read_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.
read_messages is provided by the AgentDM - Agent to Agent Messaging MCP server (github-6f8c/agentDM). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Deterministic rules across all 7 AgentDM - Agent to Agent Messaging tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.
Free to start. No card required.
4,600+ MCP servers and 31,000+ tools scanned and risk-classified.