Save an estimator/pipeline handle using sktime MLflow integration
AI agents use save_model to create or update resources in Sktime — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your Sktime environment.
This tool creates or modifies persistent model artifacts (estimator/pipeline saves to MLflow). While reversible via deletion or overwrite, it modifies the state of the model registry/storage system. The blast radius is medium because a malicious actor could poison model registries or consume storage, but cannot delete existing data or execute arbitrary code—those would be more severe categories.
From the tool's definition Tool description states 'Save an estimator/pipeline handle using sktime MLflow integration'. The verb 'Save' and the operation of persisting a model handle to storage indicates data creation/modification without deletion.
Documented attack patterns abuse exactly the kind of access save_model gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Sktime, and nothing reaches the server without passing your rules. This is the rule we recommend for save_model:
{
"version": "1",
"default": "deny",
"tools": {
"save_model": {
"limits": [
{
"counter": "save_model_rate",
"window": "minute",
"max": 30,
"scope": "grant"
}
]
}
}
}save_model stays usable, but capped — an agent stuck in a loop can't make hundreds of changes a minute. Everything else on the server is denied unless you say otherwise.
Free to start. No card required.
Save an estimator/pipeline handle using sktime MLflow integration. It is categorised as a Write tool in the Sktime MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.
Register the Sktime MCP server in PolicyLayer and add a rule for save_model: 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 Sktime. Nothing to install.
save_model is a Write tool with medium risk. Write tools should be rate-limited to prevent accidental bulk modifications.
Yes. Add a rate_limit block to the save_model 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 save_model. 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.
save_model is provided by the Sktime MCP server (sktime/sktime-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Start from Sktime, 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.
24 Sktime tools catalogued and risk-classified — across an index of 43,000+ MCP servers.