Delete one stored entry by id.
AI agents call forget to permanently remove resources in Agent Memory Bridge — typically in cleanup and lifecycle workflows. It does its job in a single call, and there is no undo.
This tool irreversibly deletes stored memory entries from the agent's knowledge base. While the blast radius is limited to the agent's own memory (not external systems), deletion of learned domain knowledge, gotchas, and decisions cannot be undone and could compromise the agent's future decision-making if misused. This fits the Destructive category as it performs an operation that cannot be reversed.
From the tool's definition 'Delete one stored entry by id' - the tool explicitly performs deletion of stored data entries, which is an irreversible operation.
Attacks that exploit this kind of access
Delete one stored entry by id. It is categorised as a Destructive tool in the Agent Memory Bridge MCP Server, which means it can permanently delete or destroy data. Block by default and require explicit approval.
Register the Agent Memory Bridge MCP server in PolicyLayer and add a rule for forget: 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 Memory Bridge. Nothing to install.
forget is a Destructive tool with critical risk. Critical-risk tools should be blocked by default and only enabled with explicit human approval.
Yes. Add a rate_limit block to the forget 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 forget. 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.
forget is provided by the Agent Memory Bridge MCP server (zzhang82/agent-memory-bridge). 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 →