Tombstone a knowledge entry (OR-Set remove). A concurrent re-post wins (add-bias).
AI agents call retract_knowledge to permanently remove resources in ATMcp — typically in cleanup and lifecycle workflows. It does its job in a single call, and there is no undo.
Tombstoning a knowledge entry is a removal/deletion operation. While the description notes 'a concurrent re-post wins (add-bias)', implying some potential recoverability under concurrent conditions, the primary action is marking an entry as removed (tombstoned), which is effectively irreversible in the absence of a concurrent re-post. This qualifies as Destructive rather than Write.
From the tool's definition Tombstone a knowledge entry (OR-Set remove)
Attacks that exploit this kind of access
Tombstone a knowledge entry (OR-Set remove). A concurrent re-post wins (add-bias). It is categorised as a Destructive tool in the ATMcp MCP Server, which means it can permanently delete or destroy data. Block by default and require explicit approval.
Register the AT MCP server in PolicyLayer and add a rule for retract_knowledge: 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 ATMcp. Nothing to install.
retract_knowledge 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 retract_knowledge 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 retract_knowledge. 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.
retract_knowledge is provided by the AT MCP server (midcheck/atmcp). 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 →