run_embedding · Model Runner MCP: Risk & Policy

WHAT IT DOES

What run_embedding does on Model Runner

AI agents invoke run_embedding to trigger actions in Model Runner. What it does depends on the arguments the agent supplies, and its effects often reach beyond the immediate call — builds kicked off, notifications sent, workflows started.

WHY IT NEEDS A POLICY

Why run_embedding needs a policy

This tool executes code/operations (model inference) against external services, making it Execute rather than Read. While embeddings are deterministic and query-like, the tool invokes remote ML model computation on user-supplied text.

From the tool's definition Tool description states 'Generate a vector embedding for text using an embedding model' with configurable provider parameter, indicating execution of external ML model inference that produces output whose nature depends on arguments.

Attacks that exploit this kind of access

Recommended rule Ready

Verdict

run_embedding Rate-limited

run_embedding stays usable, but rate-capped — a runaway agent can't fire it dozens of times a minute. Everything else on the server is denied unless you say otherwise.

View as policy code

policy.json

{
  "version": "1",
  "default": "deny",
  "tools": {
    "run_embedding": {
      "limits": [
        {
          "counter": "run_embedding_rate",
          "window": "minute",
          "max": 10,
          "scope": "grant"
        }
      ]
    }
  }
}

RATE-LIMIT THIS TOOL →

Instant setup, no code required.

PolicyLayer is an MCP gateway: every tool call is checked against your rule before it reaches Model Runner, so you can run agents against it safely.

FAQ

Questions about run_embedding

What does the run_embedding tool do? +

Generate a vector embedding for text using an embedding model. e.g., provider:. It is categorised as a Execute tool in the Model Runner MCP Server, which means it can trigger actions or run processes. Use rate limits and argument validation.

How do I enforce a policy on run_embedding? +

Register the Model Runner MCP server in PolicyLayer and add a rule for run_embedding: 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 Model Runner. Nothing to install.

What risk level is run_embedding? +

run_embedding is a Execute tool with high risk. Execute tools should be rate-limited and have argument validation enabled.

Can I rate-limit run_embedding? +

Yes. Add a rate_limit block to the run_embedding 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 run_embedding completely? +

Set action: deny in the PolicyLayer policy for run_embedding. 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 run_embedding? +

run_embedding is provided by the Model Runner MCP server (josephtandle/replicate-mcp-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.