List messages addressed to an agent. Default: unread only.
AI agents call check_inbox to retrieve information from Claude Sync without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This tool retrieves or queries existing messages from an inbox without creating, modifying, deleting, or executing anything. It is a straightforward read operation. The low severity reflects that merely listing messages has minimal blast radius—an agent could read message content but cannot alter state or trigger external operations. The confidence is high because the description is clear and unambiguous.
From the tool's definition The tool description states it 'List[s] messages addressed to an agent' with a 'Default: unread only.' The verb 'list' and action of reading messages are characteristic of retrieval operations with no side effects.
Attacks that exploit this kind of access
List messages addressed to an agent. Default: unread only. It is categorised as a Read tool in the Claude Sync MCP Server, which means it retrieves data without modifying state.
Register the Claude Sync MCP server in PolicyLayer and add a rule for check_inbox: 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 Claude Sync. Nothing to install.
check_inbox 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_inbox 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_inbox. 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_inbox is provided by the Claude Sync MCP server (the-firmament/claude-sync). 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 →