Critical Risk →

memory_delete_observations

Delete and forget specific observations, details, or information from entities in the user

How to control memory_delete_observations ↓

What memory_delete_observations does on VaultAssist

AI agents call memory_delete_observations to permanently remove resources in VaultAssist — typically in cleanup and lifecycle workflows. It does its job in a single call, and there is no undo.

Critical Risk

Why memory_delete_observations needs a policy

This tool irreversibly deletes user data (observations/details from entities), which is the defining characteristic of the Destructive category. While scoped to the user's own entity data rather than system-wide, the permanent nature of deletion and inability to recover justifies 'high' severity.

From the tool's definition Tool performs 'Delete and forget specific observations, details, or information from entities' - irreversibly removes data with no undo mechanism. The word 'delete' combined with 'forget' indicates permanent erasure.

Documented attack patterns abuse exactly the kind of access memory_delete_observations gives an agent:

How to control memory_delete_observations

PolicyLayer is an MCP gateway — it sits between your AI agents and VaultAssist, and nothing reaches the server without passing your rules. This is the rule we recommend for memory_delete_observations:

policy.json
{
  "version": "1",
  "default": "deny",
  "hide": [
    "memory_delete_observations"
  ]
}

memory_delete_observations disappears from the agent's tool list entirely, and any attempt to call it is denied. The rest of the server keeps working.

  1. Create a free account and register VaultAssist — nothing to install.
  2. Add this policy — paste it, or build it visually.
  3. Point your MCP client (Claude, Cursor, anything) at your gateway URL.
RESTRICT THIS TOOL →

Free to start. No card required.

Related tools and policies

Go deeper

Questions about memory_delete_observations

What does the memory_delete_observations tool do? +

Delete and forget specific observations, details, or information from entities in the user. It is categorised as a Destructive tool in the VaultAssist MCP Server, which means it can permanently delete or destroy data. Block by default and require explicit approval.

How do I enforce a policy on memory_delete_observations? +

Register the VaultAssist MCP server in PolicyLayer and add a rule for memory_delete_observations: 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 VaultAssist. Nothing to install.

What risk level is memory_delete_observations? +

memory_delete_observations is a Destructive tool with critical risk. Critical-risk tools should be blocked by default and only enabled with explicit human approval.

Can I rate-limit memory_delete_observations? +

Yes. Add a rate_limit block to the memory_delete_observations 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.

How do I block memory_delete_observations completely? +

Set action: deny in the PolicyLayer policy for memory_delete_observations. 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.

What MCP server provides memory_delete_observations? +

memory_delete_observations is provided by the VaultAssist MCP server (3xcaffeine/mcp-server). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.

Enforce policy on every VaultAssist tool call.

Start from VaultAssist, add the rest of your stack, and see everything your agents can call. Then put policy on all of it.

Free to start. No card required.

79 VaultAssist tools catalogued and risk-classified — across an index of 43,000+ MCP servers.

// GET IN TOUCH

Have a question or want to learn more? Send us a message.

Message sent.

We'll get back to you soon.