Cloud Providers

Shannon supports multiple AI providers beyond direct Anthropic API access. Route requests through AWS Bedrock, Google Vertex AI, or experimental multi-model providers.

Provider Options

Anthropic API

Direct API access (default, recommended)

AWS Bedrock

Route through Amazon Bedrock

Google Vertex AI

Route through Google Cloud Vertex AI

Router Mode

Experimental multi-model support

Anthropic API (Default)

Direct API access is the simplest and recommended approach.

Setup

Get API Key

Configure .env

.env

ANTHROPIC_API_KEY=your-api-key-here
CLAUDE_CODE_MAX_OUTPUT_TOKENS=64000

Run Shannon

./shannon start URL=https://example.com REPO=repo-name

OAuth Token Alternative

Use Claude Code OAuth token instead of API key:

.env

CLAUDE_CODE_OAUTH_TOKEN=your-oauth-token-here
CLAUDE_CODE_MAX_OUTPUT_TOKENS=64000

AWS Bedrock

Route requests through Amazon Bedrock instead of direct Anthropic API.

Prerequisites

AWS account with Bedrock access
Claude models enabled in your AWS region
Bedrock API keys (Bearer token)

Setup

Generate Bedrock API Key

Follow AWS Bedrock API Keys guide to generate a bearer token.

Configure .env

.env

CLAUDE_CODE_USE_BEDROCK=1
AWS_REGION=us-east-1
AWS_BEARER_TOKEN_BEDROCK=your-bearer-token

# Set models with Bedrock-specific IDs for your region
ANTHROPIC_SMALL_MODEL=us.anthropic.claude-haiku-4-5-20251001-v1:0
ANTHROPIC_MEDIUM_MODEL=us.anthropic.claude-sonnet-4-6
ANTHROPIC_LARGE_MODEL=us.anthropic.claude-opus-4-6

Run Shannon

./shannon start URL=https://example.com REPO=repo-name

Shannon automatically uses Bedrock for all requests.

Model Tiers

Shannon uses three model tiers with different Bedrock IDs per region:

Small Tier
Medium Tier
Large Tier

Used for summarization and light analysis.

ANTHROPIC_SMALL_MODEL=us.anthropic.claude-haiku-4-5-20251001-v1:0

Default: claude-haiku-4-5-20251001

Used for security analysis and vulnerability hunting.

ANTHROPIC_MEDIUM_MODEL=us.anthropic.claude-sonnet-4-6

Default: claude-sonnet-4-6

Used for deep reasoning and exploitation.

ANTHROPIC_LARGE_MODEL=us.anthropic.claude-opus-4-6

Default: claude-opus-4-6

Regional Model IDs

Bedrock model IDs vary by AWS region. Common formats:

ANTHROPIC_SMALL_MODEL=us.anthropic.claude-haiku-4-5-20251001-v1:0
ANTHROPIC_MEDIUM_MODEL=us.anthropic.claude-sonnet-4-6
ANTHROPIC_LARGE_MODEL=us.anthropic.claude-opus-4-6

Regional AvailabilityNot all Claude models are available in all AWS regions. Check the AWS Bedrock Model Catalog for your region’s supported models.

Configuration Reference

CLAUDE_CODE_USE_BEDROCK

string

Enable Bedrock mode (set to 1)

AWS_REGION

string

required

AWS region for Bedrock (e.g., us-east-1, us-west-2)

AWS_BEARER_TOKEN_BEDROCK

string

required

Bedrock API bearer token

ANTHROPIC_SMALL_MODEL

string

required

Bedrock model ID for small tier

ANTHROPIC_MEDIUM_MODEL

string

required

Bedrock model ID for medium tier

ANTHROPIC_LARGE_MODEL

string

required

Bedrock model ID for large tier

Google Vertex AI

Route requests through Google Cloud Vertex AI.

Prerequisites

Google Cloud Platform project
Vertex AI API enabled
Service account with roles/aiplatform.user role

Setup

Create Service Account

In the GCP Console:

Create a new service account
Grant roles/aiplatform.user role
Create and download a JSON key file

Place Key File

mkdir -p ./credentials
cp /path/to/your-sa-key.json ./credentials/gcp-sa-key.json

The ./credentials/ directory is mounted into the Docker container.

Configure .env

.env

CLAUDE_CODE_USE_VERTEX=1
CLOUD_ML_REGION=us-east5
ANTHROPIC_VERTEX_PROJECT_ID=your-gcp-project-id
GOOGLE_APPLICATION_CREDENTIALS=./credentials/gcp-sa-key.json

# Set models with Vertex AI model IDs
ANTHROPIC_SMALL_MODEL=claude-haiku-4-5@20251001
ANTHROPIC_MEDIUM_MODEL=claude-sonnet-4-6
ANTHROPIC_LARGE_MODEL=claude-opus-4-6

Run Shannon

./shannon start URL=https://example.com REPO=repo-name

Regional Endpoints

Vertex AI supports global and regional endpoints:

Global Endpoint
Regional Endpoint

CLOUD_ML_REGION=global

Some models may not be available on global endpoints. Use regional endpoints for full model availability.

CLOUD_ML_REGION=us-east5

Recommended regions:

us-east5 (Virginia)
us-central1 (Iowa)
europe-west1 (Belgium)

Check the Vertex AI Model Garden for region availability.

Model IDs

Vertex AI uses different model ID formats than Anthropic or Bedrock:

ANTHROPIC_SMALL_MODEL=claude-haiku-4-5@20251001
ANTHROPIC_MEDIUM_MODEL=claude-sonnet-4-6
ANTHROPIC_LARGE_MODEL=claude-opus-4-6

Configuration Reference

CLAUDE_CODE_USE_VERTEX

string

Enable Vertex AI mode (set to 1)

CLOUD_ML_REGION

string

required

GCP region or global for global endpoint

ANTHROPIC_VERTEX_PROJECT_ID

string

required

Your GCP project ID

GOOGLE_APPLICATION_CREDENTIALS

string

required

Path to service account JSON key file (must be in ./credentials/)

ANTHROPIC_SMALL_MODEL

string

required

Vertex AI model ID for small tier

ANTHROPIC_MEDIUM_MODEL

string

required

Vertex AI model ID for medium tier

ANTHROPIC_LARGE_MODEL

string

required

Vertex AI model ID for large tier

Router Mode (Experimental)

Experimental FeatureRouter mode is experimental and unsupported. It’s intended for model experimentation only. Output quality varies significantly by model and may fail early phases.Shannon is optimized for Anthropic Claude models. Alternative providers may produce inconsistent results.

Overview

Router mode uses claude-code-router to route requests through alternative AI providers:

OpenAI: GPT-5.2, GPT-5-mini
OpenRouter: Google Gemini 3 Flash Preview

Setup

Configure Provider

Add your provider API key to .env:

OpenAI
OpenRouter

.env

OPENAI_API_KEY=sk-...
ROUTER_DEFAULT=openai,gpt-5.2

.env

OPENROUTER_API_KEY=sk-or-...
ROUTER_DEFAULT=openrouter,google/gemini-3-flash-preview

Run with ROUTER=true

./shannon start URL=https://example.com REPO=repo-name ROUTER=true

The ROUTER=true flag starts the router container and routes all requests through it.

Available Models

OpenAI
OpenRouter

ROUTER_DEFAULT=openai,gpt-5.2

Models:

gpt-5.2
gpt-5-mini

ROUTER_DEFAULT=openrouter,google/gemini-3-flash-preview

Models:

google/gemini-3-flash-preview

Router Configuration

Router settings are defined in configs/router-config.json:

configs/router-config.json

{
  "HOST": "0.0.0.0",
  "APIKEY": "shannon-router-key",
  "Providers": [
    {
      "name": "openai",
      "api_base_url": "https://api.openai.com/v1/chat/completions",
      "api_key": "$OPENAI_API_KEY",
      "models": ["gpt-5.2", "gpt-5-mini"]
    },
    {
      "name": "openrouter",
      "api_base_url": "https://openrouter.ai/api/v1/chat/completions",
      "api_key": "$OPENROUTER_API_KEY",
      "models": ["google/gemini-3-flash-preview"]
    }
  ],
  "Router": {
    "default": "$ROUTER_DEFAULT"
  }
}

Limitations

Inconsistent results

Alternative models may:

Fail reconnaissance phase
Miss vulnerability patterns
Generate invalid exploits
Produce hallucinated findings

No official support

Router mode is provided as-is for experimentation. Issues specific to alternative models will not be addressed.

Performance varies

Model performance depends heavily on:

Provider API reliability
Model capabilities (reasoning, tool use)
Request routing latency

Provider Comparison

Anthropic API

Recommended

Pros:

Simplest setup
Best performance
Direct support
No intermediary services

Cons:

Requires Anthropic account

AWS Bedrock

Pros:

Use existing AWS infrastructure
Unified billing with AWS
Enterprise compliance features

Cons:

More complex setup
Regional model availability varies
Requires AWS account and Bedrock access

Google Vertex AI

Pros:

Use existing GCP infrastructure
Unified billing with GCP
Service account security model

Cons:

More complex setup
Regional model availability varies
Requires GCP project

Router Mode

Experimental

Pros:

Experiment with alternative models
Test GPT or Gemini performance

Cons:

Unsupported
Inconsistent results
May fail early phases
Complex debugging

Troubleshooting

Bedrock: Missing model tier variables

Error:

ERROR: Bedrock mode requires the following env vars in .env:
AWS_REGION ANTHROPIC_SMALL_MODEL ANTHROPIC_MEDIUM_MODEL ANTHROPIC_LARGE_MODEL

Solution: Set all three model tier variables with Bedrock-specific IDs for your region.

Vertex AI: Service account key not found

Error:

ERROR: Service account key file not found: ./credentials/gcp-sa-key.json

Solution: Ensure the key file is placed in ./credentials/ directory (not elsewhere).

Vertex AI: Permission denied

Error:

ERROR: Permission denied on Vertex AI API

Solution: Verify service account has roles/aiplatform.user role in GCP IAM.

Router: No provider API key

Error:

WARNING: No provider API key set (OPENAI_API_KEY or OPENROUTER_API_KEY)

Solution: Set the appropriate provider API key in .env before running with ROUTER=true.

Router: Failed early phases

Issue: Reconnaissance or analysis phases fail with alternative models.Solution: Router mode is experimental. Switch back to Anthropic Claude models for reliable results.

Configuration

Complete YAML configuration reference

Authentication

Setup login flows and 2FA

Get Started

Core Concepts

Guides

CLI Reference

Cloud Providers

Provider Options

Anthropic API

AWS Bedrock

Google Vertex AI

Router Mode

Anthropic API (Default)

Setup

OAuth Token Alternative

AWS Bedrock

Prerequisites

Setup

Model Tiers

Regional Model IDs

Configuration Reference

Google Vertex AI

Prerequisites

Setup

Regional Endpoints

Model IDs

Configuration Reference

Router Mode (Experimental)

Overview

Setup

Available Models

Router Configuration

Limitations

Provider Comparison

Troubleshooting

Configuration

Authentication

Build docs developers (and LLMs) love

Get Started

Core Concepts

Guides

CLI Reference

​Provider Options

Anthropic API

AWS Bedrock

Google Vertex AI

Router Mode

​Anthropic API (Default)

​Setup

​OAuth Token Alternative

​AWS Bedrock

​Prerequisites

​Setup

​Model Tiers

​Regional Model IDs

​Configuration Reference

​Google Vertex AI

​Prerequisites

​Setup

​Regional Endpoints

​Model IDs

​Configuration Reference

​Router Mode (Experimental)

​Overview

​Setup

​Available Models

​Router Configuration

​Limitations

​Provider Comparison

​Troubleshooting

​Related Resources

Configuration

Authentication

Build docs developers (and LLMs) love

Provider Options

Anthropic API (Default)

Setup

OAuth Token Alternative

AWS Bedrock

Prerequisites

Setup

Model Tiers

Regional Model IDs

Configuration Reference

Google Vertex AI

Prerequisites

Setup

Regional Endpoints

Model IDs

Configuration Reference

Router Mode (Experimental)

Overview

Setup

Available Models

Router Configuration

Limitations

Provider Comparison

Troubleshooting

Related Resources