// SCAN REPORT

Replicate MCP Server

13 tools. 4 can modify or destroy data without limits.

1 destructive tool with no built-in limits. Policy required.

Last updated: 31 Jul 2026

4 can modify or destroy data

9 read-only

13 tools total

Community server · catalogue entry checked 31/07/2026

How to control Replicate MCP Server ↓

TOOL MAP

What Replicate MCP Server exposes to your agents

Read (9) Write / Execute (3) Destructive / Financial (1)

Running more than one server? Check your whole stack →

MOST DANGEROUS TOOLS

Critical Risk

The most dangerous Replicate MCP Server tools

Destructive · Medium clear_image_cache The tool performs cache clearing, which permanently removes stored image data. Execute · Medium cancel_prediction Cancelling a running prediction is an external operation that terminates an in-progress computation on Replicate's infrastructure. Execute · High create_and_poll_prediction This tool executes an AI model prediction on an external platform (Replicate) and waits for the result. Execute · High create_prediction Creating a prediction executes an AI model on external compute infrastructure.

4 of Replicate MCP Server's 13 tools can modify, destroy, or commit something on every call — and an agent calls them with no built-in limits.

POLICY

How to control Replicate MCP Server

PolicyLayer is an MCP gateway — it sits between your AI agents and Replicate MCP Server, and nothing reaches the server without passing your rules. These are the rules we recommend:

Deny destructive operations

{
  "clear_image_cache": {
    "deny_if": [
      {
        "conditions": [],
        "on_deny": "Blocked by default. Requires approval."
      }
    ]
  }
}

Destructive tools should never be available to autonomous agents without human approval.

Cap read operations

{
  "get_collection": {
    "limits": [
      {
        "counter": "get_collection_per_minute",
        "window": "minute",
        "max": 60,
        "scope": "grant"
      }
    ]
  }
}

Controls API costs and prevents retry loops from exhausting upstream rate limits.

Create a free account and register Replicate MCP Server — nothing to install.
Add these rules — paste them, or build them visually. Tune the limits to your setup.
Point your MCP client (Claude, Cursor, anything) at your gateway URL.

ENFORCE POLICY ON REPLICATE →

Instant setup, no code required.

FULL CATALOGUE

All 13 Replicate MCP Server tools

DESTRUCTIVE 1 tools

Destructive clear_image_cache Clear the image viewer cache

EXECUTE 3 tools

Execute cancel_prediction Cancel a running prediction Execute create_and_poll_prediction Create a new prediction and wait until it Execute create_prediction Create a new prediction using either a model version (for community models) or model name (for official models

READ 9 tools

Read get_collection Get details of a specific collection Read get_image_cache_stats Get statistics about the image cache Read get_model Get details of a specific model including available versions Read get_prediction Get details about a specific prediction Read list_collections List available model collections Read list_models List available models with optional filtering Read list_predictions List recent predictions Read search_models Search for models using semantic search Read view_image Display an image in the system

EXPLORE

Related servers

Other MCP servers with similar tools — same risk classification, starter policies for each.

FAQ

Questions about Replicate MCP Server

Can an AI agent delete data through the Replicate MCP Server MCP server? +

Yes. The Replicate MCP Server server exposes 1 destructive tools including clear_image_cache. These permanently remove resources with no undo. PolicyLayer blocks destructive tools by default so they never reach the upstream server.

How many tools does the Replicate MCP Server MCP server expose? +

13 tools across 3 categories: Destructive, Read, Write. 9 are read-only. 4 can modify, create, or delete data.

How do I enforce a policy on Replicate MCP Server? +

Register the Replicate MCP Server MCP server in PolicyLayer, apply the suggested rules above (adjust the limits to your use case), and point your AI client at the PolicyLayer proxy URL instead of the server directly. Your agents keep the same tools; PolicyLayer evaluates every call against policy before it executes. Nothing to install, live in minutes.

Enforce policy on every Replicate MCP Server tool call.

Deterministic rules across all 13 Replicate MCP Server tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.

ENFORCE POLICY ON REPLICATE →

Instant setup, no code required.

13 Replicate MCP Server tools catalogued and risk-classified — across an index of 46,500+ MCP servers.

// WHERE THIS COMES FROM

These policies come from Replicate MCP Server's registry record.

The record behind this page: verified identity, auth posture, risk grade, every tool classified, recommended policy — re-checked continuously.

Teams ship this data inside their own products. See what a licence covers →