Snowflake Native App

For teams already on Snowflake — install directly from the Marketplace with full data sovereignty.

Overview

Prometheux is available as a Snowflake Native App that runs entirely within your Snowflake environment. All processing happens inside your account, and the app leverages Snowflake’s compute, security, and access controls natively. Key advantages:

Data sovereignty — data is processed within Snowflake
Snowflake-native security — leverages Snowflake’s access controls
Elastic compute — scales with Snowflake warehouses and compute pools
Marketplace distribution — easy installation and updates

Prerequisites

Snowflake Account: Any tier (Standard, Enterprise, Business Critical)
Privileges: ACCOUNTADMIN role for initial setup, or specific grants for non-admin users
Compute Resources: Access to create or use compute pools
Warehouse: Any warehouse for database operations

The Snowflake Native Apps framework automatically handles privilege granting for application database access, service management, snapshot operations, and configuration storage.

Quick Start

Install directly from the Snowflake Marketplace, then launch from the Snowsight UI:

Navigate to Snowsight — Go to your Snowflake web interface
Find the App — Navigate to Data > Apps > Prometheux
Launch — Click the “Launch App” button
Initialize — The app will auto-configure and provide a launch URL

The platform will automatically create required databases and schemas, set up default configuration, and provide a platform access URL.

View Configuration

USE DATABASE prometheux;
USE SCHEMA v1;

CALL get_app_configuration();

Warehouse Configuration

Prometheux automatically creates a dedicated warehouse called PROMETHEUX_WAREHOUSE for read/write operations on your Snowflake databases.

Setting	Value
Size	SMALL
Auto-suspend	180 seconds (3 minutes)
Auto-resume	Enabled
Approximate cost	$0.0023/hour when active

Prometheux minimizes warehouse usage costs by storing application data in an internal PostgreSQL database within the compute pool, only using the Snowflake warehouse for operations on your Snowflake databases.

To use an existing warehouse alongside the app warehouse:

GRANT USAGE ON WAREHOUSE YOUR_EXISTING_WH TO APPLICATION prometheux;

Essential Procedures

Core Platform Management

USE DATABASE prometheux;
USE SCHEMA v1;

-- Initialize/restart platform with stored configuration
CALL init();

-- Get platform access URL
CALL platform_url();

-- Stop platform (automatically creates snapshot)
CALL stop_platform();

Configuration Management

-- View current configuration
CALL get_app_configuration();

-- Set compute pool
CALL set_compute_pool('prometheux_compute_pool');

-- Add external access integrations (comma-separated list)
CALL add_external_access_integrations('integration1,integration2');

-- Remove external access integrations
CALL remove_external_access_integrations('integration1');

Monitoring

-- Check snapshot status
CALL check_snapshots();

View application logs via Snowflake’s telemetry events:

SELECT
    TIMESTAMP                      AS time_utc,
    VALUE::string                  AS message
FROM SNOWFLAKE.TELEMETRY.EVENTS
WHERE RECORD_TYPE = 'LOG'
  AND (
        RESOURCE_ATTRIBUTES['snow.application.name'] = 'PROMETHEUX'
        OR RESOURCE_ATTRIBUTES['snow.database.name'] = 'PROMETHEUX'
      )
ORDER BY time_utc DESC;

Uploading Data

Snowflake’s shared file system for Native Apps does not support filenames with spaces. Use underscores (_) or hyphens (-) instead.

Via Snowsight UI: Navigate to Catalog > Database Explorer > PROMETHEUX > APP_PUBLIC > STAGES > USER_UPLOADS and use the upload button. Via SQL:

USE DATABASE prometheux;
USE SCHEMA app_public;

PUT file://path/to/your.csv @user_uploads AUTO_COMPRESS=FALSE;

The system automatically transfers your file to the internal block volume and makes it available to all back-end components.

Granting Access to Existing Databases

Prometheux needs different levels of access depending on your use case.

Read-Only Access

USE ROLE ACCOUNTADMIN;

GRANT USAGE ON DATABASE YOUR_DATABASE TO APPLICATION prometheux;
GRANT USAGE ON SCHEMA YOUR_DATABASE.YOUR_SCHEMA TO APPLICATION prometheux;
GRANT SELECT ON ALL TABLES IN SCHEMA YOUR_DATABASE.YOUR_SCHEMA TO APPLICATION prometheux;
GRANT SELECT ON ALL VIEWS IN SCHEMA YOUR_DATABASE.YOUR_SCHEMA TO APPLICATION prometheux;

SHOW GRANTS TO APPLICATION prometheux;

Read-Write Access

USE ROLE ACCOUNTADMIN;

-- Database and schema access
GRANT USAGE ON DATABASE YOUR_DATABASE TO APPLICATION prometheux;
GRANT USAGE ON SCHEMA YOUR_DATABASE.YOUR_SCHEMA TO APPLICATION prometheux;

-- Schema-level DDL privileges
GRANT CREATE TABLE ON SCHEMA YOUR_DATABASE.YOUR_SCHEMA TO APPLICATION prometheux;
GRANT CREATE VIEW ON SCHEMA YOUR_DATABASE.YOUR_SCHEMA TO APPLICATION prometheux;
GRANT CREATE SEQUENCE ON SCHEMA YOUR_DATABASE.YOUR_SCHEMA TO APPLICATION prometheux;
GRANT CREATE FUNCTION ON SCHEMA YOUR_DATABASE.YOUR_SCHEMA TO APPLICATION prometheux;
GRANT CREATE PROCEDURE ON SCHEMA YOUR_DATABASE.YOUR_SCHEMA TO APPLICATION prometheux;

-- Table-level DML privileges
GRANT SELECT, INSERT, UPDATE, DELETE ON ALL TABLES IN SCHEMA YOUR_DATABASE.YOUR_SCHEMA TO APPLICATION prometheux;

-- View, sequence, function, and procedure access
GRANT SELECT ON ALL VIEWS IN SCHEMA YOUR_DATABASE.YOUR_SCHEMA TO APPLICATION prometheux;
GRANT USAGE ON ALL SEQUENCES IN SCHEMA YOUR_DATABASE.YOUR_SCHEMA TO APPLICATION prometheux;
GRANT USAGE ON ALL FUNCTIONS IN SCHEMA YOUR_DATABASE.YOUR_SCHEMA TO APPLICATION prometheux;
GRANT USAGE ON ALL PROCEDURES IN SCHEMA YOUR_DATABASE.YOUR_SCHEMA TO APPLICATION prometheux;

SHOW GRANTS TO APPLICATION prometheux;

Granular Access (Specific Tables)

USE ROLE ACCOUNTADMIN;

GRANT USAGE ON DATABASE YOUR_DATABASE TO APPLICATION prometheux;
GRANT USAGE ON SCHEMA YOUR_DATABASE.YOUR_SCHEMA TO APPLICATION prometheux;

GRANT SELECT ON TABLE YOUR_DATABASE.YOUR_SCHEMA.CUSTOMERS TO APPLICATION prometheux;
GRANT SELECT ON TABLE YOUR_DATABASE.YOUR_SCHEMA.ORDERS TO APPLICATION prometheux;
GRANT SELECT, INSERT, UPDATE, DELETE ON TABLE YOUR_DATABASE.YOUR_SCHEMA.ANALYSIS_RESULTS TO APPLICATION prometheux;

SHOW GRANTS TO APPLICATION prometheux;

External Access Configuration

To enable Prometheux to connect to external services, create and configure external access integrations in Snowflake. Create the integrations first, then add them to the app. Once integrations are created, add them to the app and restart:

USE DATABASE prometheux;
USE SCHEMA v1;

CALL add_external_access_integrations('prometheux_outbound,azure_openai_outbound');
CALL init();

Modifying Compute Pool Settings

By default, the platform starts with 1 node using the CPU_X64_M instance family. You can change this based on your performance requirements.

Available Instance Families

CPU-Optimized:

Instance Family	vCPUs	Memory	Cost
`CPU_X64_M` (default)	6	28 GB	$0.59/hr (0.22 credits/hr)
`CPU_X64_SL`	14	54 GB	$1.11/hr (0.41 credits/hr)
`CPU_X64_L`	28	116 GB	$2.24/hr (0.83 credits/hr)

High-Memory:

Instance Family	vCPUs	Memory	Cost	Availability
`HIGHMEM_X64_S`	6	58 GB	$0.76/hr (0.28 credits/hr)	All clouds
`HIGHMEM_X64_M`	28	240–244 GB	$3.00/hr (1.11 credits/hr)	All clouds
`HIGHMEM_X64_SL`	92	654 GB	$7.91/hr (2.93 credits/hr)	Azure/GCP only
`HIGHMEM_X64_L`	124	984 GB	$11.99/hr (4.44 credits/hr)	AWS only

Changing Compute Pool

USE ROLE ACCOUNTADMIN;

CREATE COMPUTE POOL high_performance_pool FOR APPLICATION prometheux
    MIN_NODES = 1
    MAX_NODES = 1
    INSTANCE_FAMILY = CPU_X64_L
    AUTO_RESUME = TRUE;

GRANT USAGE ON COMPUTE POOL high_performance_pool TO APPLICATION prometheux;

USE DATABASE prometheux;
USE SCHEMA v1;

CALL set_compute_pool('high_performance_pool');
CALL init();

Snowflake Cortex AI Assistant

Prometheux can use Snowflake Cortex as its built-in AI assistant, so all LLM inference stays within your Snowflake account — no external API keys required.

Enable Cortex access

When installing or upgrading the app, approve the IMPORTED PRIVILEGES ON DATABASE SNOWFLAKE permission in the app’s Permissions tab. This grants the app access to Cortex LLM functions. If your Snowflake region has limited model availability, also run the following as ACCOUNTADMIN:

USE ROLE ACCOUNTADMIN;

-- Enable cross-region inference so all Cortex models are available
-- Use 'AWS_EU' to keep data within EU, or 'ANY_REGION' for full model access
ALTER ACCOUNT SET CORTEX_ENABLED_CROSS_REGION = 'ANY_REGION';

Select Cortex in the app

Open the Prometheux app and go to Settings > Chat Model Configuration
Select snowflake cortex as the LLM provider
Choose a model from the dropdown (populated dynamically from your account)
Click Save

The AI assistant will now use Snowflake Cortex for all responses. Available models depend on your region and cross-region settings — see Snowflake Cortex model availability for details.

No external access neededUnlike OpenAI or Azure providers, Cortex requires no external access integration or API keys. All inference is handled natively within Snowflake.

Cortex Agent Integration (MCP)

The native app includes a built-in MCP server that is automatically registered when the platform starts. Cortex Agents and Snowflake Intelligence can discover and invoke Prometheux tools (list concepts, run reasoning, query knowledge graphs) without any additional setup.

Verify the MCP server

After CALL init() completes, confirm the MCP server is registered:

SHOW CUSTOM MCP SERVERS IN APPLICATION prometheux;
DESC CUSTOM MCP SERVER prometheux.app_public.prometheux_mcp;

Use with a Cortex Agent

Create an agent that references the native app’s MCP server:

CREATE AGENT my_prometheux_agent
  FROM SPECIFICATION $$
models:
  orchestration: auto
instructions:
  response: "Use Prometheux tools to answer questions about knowledge graphs."
tools: []
mcp_servers:
  - server_spec:
      name: "PROMETHEUX.APP_PUBLIC.PROMETHEUX_MCP"
$$;

GRANT USAGE ON AGENT my_prometheux_agent TO ROLE my_role;

The agent will automatically discover Prometheux tools via tools/list and invoke them as needed.

Cortex CodeThe built-in MCP server works with Cortex Agents and Snowflake Intelligence. For Cortex Code (Snowflake’s AI coding assistant), use the external MCP connector pointing at the cloud-hosted Prometheux platform instead.

Available ToolsSee the MCP overview page for the complete list of tools exposed via MCP.

Next Steps

Vadalog Reference — Understand the core building blocks
Python SDK — Connect to Prometheux programmatically
MCP Server — Integrate with Claude and other AI assistants

For feedback or enterprise use cases, contact the Prometheux team at support@prometheux.co.uk.

Introduction

Get Started

Platform

Integrations

Examples

Vadalog Reference

Resources

Snowflake Native App

Overview

Prerequisites

Quick Start

View Configuration

Warehouse Configuration

Essential Procedures

Core Platform Management

Configuration Management

Monitoring

Uploading Data

Granting Access to Existing Databases

Read-Only Access

Read-Write Access

Granular Access (Specific Tables)

External Access Configuration

Modifying Compute Pool Settings

Available Instance Families

Changing Compute Pool

Snowflake Cortex AI Assistant

Enable Cortex access

Select Cortex in the app

Cortex Agent Integration (MCP)

Verify the MCP server

Use with a Cortex Agent

Next Steps

​Overview

​Prerequisites

​Quick Start

​View Configuration

​Warehouse Configuration

​Essential Procedures

​Core Platform Management

​Configuration Management

​Monitoring

​Uploading Data

​Granting Access to Existing Databases

​Read-Only Access

​Read-Write Access

​Granular Access (Specific Tables)

​External Access Configuration

​Modifying Compute Pool Settings

​Available Instance Families

​Changing Compute Pool

​Snowflake Cortex AI Assistant

​Enable Cortex access

​Select Cortex in the app

​Cortex Agent Integration (MCP)

​Verify the MCP server

​Use with a Cortex Agent

​Next Steps

Overview

Prerequisites

Quick Start

View Configuration

Warehouse Configuration

Essential Procedures

Core Platform Management

Configuration Management

Monitoring

Uploading Data

Granting Access to Existing Databases

Read-Only Access

Read-Write Access

Granular Access (Specific Tables)

External Access Configuration

Modifying Compute Pool Settings

Available Instance Families

Changing Compute Pool

Snowflake Cortex AI Assistant

Enable Cortex access

Select Cortex in the app

Cortex Agent Integration (MCP)

Verify the MCP server

Use with a Cortex Agent

Next Steps