Delete a model from Ollama
AI agents call ollama_delete to permanently remove resources in ML Lab MCP — typically in cleanup and lifecycle workflows. It does its job in a single call, and there is no undo.
This tool permanently removes trained or downloaded ML models without recovery option. While the blast radius is primarily limited to ML artifacts (not production data or financial systems), the loss of a trained model represents significant irreversible damage in an ML engineering context. This qualifies as Destructive rather than Write because deletion cannot be undone—the model must be retrained or re-downloaded.
From the tool's definition Tool name 'ollama_delete' with description 'Delete a model from Ollama' explicitly performs irreversible deletion of ML models stored in the Ollama system.
Attacks that exploit this kind of access
Delete a model from Ollama. It is categorised as a Destructive tool in the ML Lab MCP MCP Server, which means it can permanently delete or destroy data. Block by default and require explicit approval.
Register the ML Lab MCP server in PolicyLayer and add a rule for ollama_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 ML Lab MCP. Nothing to install.
ollama_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 ollama_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 ollama_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.
ollama_delete is provided by the ML Lab MCP server (pushpullcommitpush/ml-mcp). 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 →