AI agents call memory_proposals to retrieve information from Loom without modifying anything — typically the context-gathering step in research, monitoring, and reporting workflows, before the agent takes action elsewhere.
This is a read-only operation that retrieves and displays existing data (pending proposals) without creating, modifying, or deleting anything. The blast radius of misuse is minimal—an agent could only obtain visibility into pending proposals, which poses no risk of unintended data loss, execution, or financial harm.
From the tool's definition Tool 'memory_proposals' lists pending proposals from a queue; the verb 'list' and the action of querying a queue without modification indicates retrieval only. No side effects or data mutation.
Attacks that exploit this kind of access
List all pending proposals in the capture-propose queue, newest first. It is categorised as a Read tool in the Loom MCP Server, which means it retrieves data without modifying state.
Register the Loom MCP server in PolicyLayer and add a rule for memory_proposals: 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 Loom. Nothing to install.
memory_proposals 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 memory_proposals 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 memory_proposals. 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.
memory_proposals is provided by the Loom MCP server (jbarket/loom). 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 →