Remove a key-value pair from persistent local memory.
AI agents call kv_delete to permanently remove resources in AgentOS — typically in cleanup and lifecycle workflows. It does its job in a single call, and there is no undo.
This tool permanently deletes data from persistent memory without any undo capability. While the blast radius is scoped to key-value pairs (not a full database wipe), deletion of agent memory state could break workflows, lose critical context, or corrupt agent decision-making. This qualifies as Destructive rather than Write because the operation cannot be reversed and represents loss of state.
From the tool's definition Tool name 'kv_delete' and description 'Remove a key-value pair from persistent local memory' explicitly indicate irreversible deletion of stored data.
Attacks that exploit this kind of access
Remove a key-value pair from persistent local memory. It is categorised as a Destructive tool in the AgentOS MCP Server, which means it can permanently delete or destroy data. Block by default and require explicit approval.
Register the AgentOS MCP server in PolicyLayer and add a rule for kv_delete: 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 AgentOS. Nothing to install.
kv_delete 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 kv_delete 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 kv_delete. 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.
kv_delete is provided by the AgentOS MCP server (netflypsb/agentos). 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 →