Install Scanning module in your Cloud

Overview

Red Teaming Modules involves sending malicious agentic requests to your (staging) server. By default, these malicious scanning requests are sent from the Red Teaming module installed within Akto Cloud.

There could be multiple reasons why you'd want to install probing module within your Cloud.

1. Whitelisting Akto's IP in Security Group or WAF isn't an option

2. The staging server isn't reachable from public domain

3. The WAF would block most requests (or block Akto's IP)

4. The Agentic component domain isn't resolvable from public domain

5. The Agentic component is completely internal

Copy the JWT Token

1. Login to Akto dashboard at app.akto.io

2. Go to Connectors in the left nav.

3. Open the Setup Guardrail card and copy your token (also referred to as the database abstractor token in later steps).

You then have to use a Linux VM or a Helm chart to install Akto AI Red Teaming module in your cloud.

Setup with Helm Chart

1

Add the Akto Helm Repository

Run the following commands to add and update the Akto Helm repo:

helm repo add akto https://akto-api-security.github.io/helm-charts
helm repo update

2

Install the Chart

Replace <key> with your Anthropic API Key and <token> with the Database Abstractor Token copied from Copy the JWT Token, then pick one of the following installation options.

1. Directly using the key and token

   helm install akto-mini-testing akto/akto-mini-testing \
     --set testing.agentTesting.enabled=true \
     --set testing.agentTesting.env.anthropicApiKey="<key>" \
     --set testing.aktoApiSecurityTesting.env.databaseAbstractorToken="<token>"

2. Storing the key and token in Kubernetes Secrets

   helm install akto-mini-testing akto/akto-mini-testing \
     --set testing.agentTesting.enabled=true \
     --set testing.agentTesting.env.useSecretsForAnthropicApiKey=true \
     --set testing.agentTesting.env.anthropicApiKeySecrets.anthropicSecretKey="<key>" \
     --set testing.aktoApiSecurityTesting.env.useSecretsForDatabaseAbstractorToken=true \
     --set testing.aktoApiSecurityTesting.env.databaseAbstractorTokenSecrets.token="<token>"

3. Bringing your own existing Secrets

   helm install akto-mini-testing akto/akto-mini-testing \
     --set testing.agentTesting.enabled=true \
     --set testing.agentTesting.env.useSecretsForAnthropicApiKey=true \
     --set testing.agentTesting.env.anthropicApiKeySecrets.existingSecret=<my-anthropic-secret> \
     --set testing.aktoApiSecurityTesting.env.useSecretsForDatabaseAbstractorToken=true \
     --set testing.aktoApiSecurityTesting.env.databaseAbstractorTokenSecrets.existingSecret=<my-db-token-secret>

   Each existing secret must be of type Opaque and store its value under a key named token.

Anthropic API Key Required

You must provide your actual Anthropic API Key and the Database Abstractor Service Token (JWT), either directly or via a Kubernetes Secret.

Setup Linux VM

1

Provision a New VM

Minimum recommended configuration:

• Platform: Amazon Linux 2023

• CPU: 2 vCPUs

• Memory: 4 GB RAM

• Disk: 20 GB

Don’t use burstable instances.

• Network:

  • Private subnet

  • connectivity to internet (typically via NAT)

  • connectivity to your staging service

• Security groups

  • Inbound - Open only port 22 for SSH

  • Outbound - Open all

2

SSH into the VM

1. SSH into this new instance in your Cloud

2. Run the following command:

   sudo su -

3

Install Docker & Docker Compose

Install the docker and docker-compose.

4

Create the Environment File

1. Create:

   nano docker-agentic-testing.env

2. Add the following common configuration, then append the block for your chosen LLM provider below. The scanning module supports Anthropic, Azure OpenAI, and Google Vertex AI as the LLM backing the red teaming engine.

   NODE_ENV=dev
   PORT=5500
   AGENTIC_MODE=false
   NODE_TLS_REJECT_UNAUTHORIZED=0
   AKTO_UTILITY_SERVER=http://akto-api-security-testing:8001
   USE_SESSION_MANAGEMENT=true

Anthropic

ANTHROPIC_API_KEY=<key>

Anthropic API Key Required

You must replace your actual Anthropic API Key in the env file.

Azure OpenAI

LLM_PROVIDER=azure
AZURE_OPENAI_ENDPOINT=<azure-openai-endpoint>
AZURE_OPENAI_API_KEY=<azure-openai-api-key>
AZURE_OPENAI_MODEL=<azure-openai-model>

Azure OpenAI Credentials Required

You must replace AZURE_OPENAI_ENDPOINT, AZURE_OPENAI_API_KEY, and AZURE_OPENAI_MODEL with your actual Azure OpenAI deployment details.

Google Vertex AI

LLM_PROVIDER=vertex
VERTEX_PROJECT_ID=<gcp-project-id>
VERTEX_LOCATION=<gcp-region>
VERTEX_ENDPOINT_ID=<vertex-endpoint-id>
VERTEX_ENDPOINT_DOMAIN=<vertex-endpoint-domain>
GOOGLE_CREDENTIALS_JSON=<google-service-account-credentials-json>

Vertex AI Credentials Required

You must replace VERTEX_PROJECT_ID, VERTEX_LOCATION, VERTEX_ENDPOINT_ID, VERTEX_ENDPOINT_DOMAIN, and GOOGLE_CREDENTIALS_JSON with your actual Vertex AI project and service account details.

You can also reference the original template is here.

5

Create the Docker Compose File

1. Create:

   nano docker-compose-mini-testing-agentic.yml

Important Requirements

• You must replace <your-database-abstractor-token> with the actual Database Abstractor Service Token (JWT) copied from the Step 3 of Copy the JWT Token.

• Ensure both files below are in the same directory:

  • docker-compose-mini-testing-agentic.yml

  • docker-agentic-testing.env

1. Add the following configuration:

version: '3.8'
services:
  agent-testing:
    container_name: agent-testing
    image: public.ecr.aws/aktosecurity/akto-agentic-testing:latest
    ports:
      - "5500:5500"
    env_file:
      - ./docker-agentic-testing.env
    restart: always

  akto-api-security-testing:
    image: public.ecr.aws/aktosecurity/akto-api-security-mini-testing:latest
    container_name: akto-api-security-testing
    environment:
      RUNTIME_MODE: hybrid
      DATABASE_ABSTRACTOR_SERVICE_TOKEN: <token>
      PUPPETEER_REPLAY_SERVICE_URL: "http://akto-puppeteer-replay:3000"
      MINI_TESTING_NAME: "akto-testing-module"
      AGENT_BASE_URL: "http://agent-testing:5500"
    ports:
      - "8001:8001"
    restart: always

  akto-api-security-puppeteer-replay:
    image: public.ecr.aws/aktosecurity/akto-puppeteer-replay:latest
    container_name: akto-puppeteer-replay
    ports:
      - "3000:3000"
    environment:
      NODE_ENV: production
    restart: always

  watchtower:
    image: containrrr/watchtower
    restart: always
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock
    environment:
      WATCHTOWER_CLEANUP: true
      WATCHTOWER_POLL_INTERVAL: 1800
    labels:
      com.centurylinklabs.watchtower.enable: "false"

For the reference,the original template is here.

6

Start the Scanning Module

• Run:

  docker-compose -f docker-compose-mini-testing-agentic.yml up -d

• Run the following command to ensure Docker starts up in case of instance restarts:

  systemctl enable /usr/lib/systemd/system/docker.service

Get Support for your Akto setup

There are multiple ways to request support from Akto. We are 24X7 available on the following:

1. In-app intercom support. Message us with your query on intercom in Akto dashboard and someone will reply.

2. Join our discord channel for community support.

3. Contact [e-mail] for email support.

4. Contact us here.