https://github.com/scttfrdmn/genkit-aws
AWS plugins for Google's GenKit framework - add AWS Bedrock models and CloudWatch monitoring to your existing GenKit applications
https://github.com/scttfrdmn/genkit-aws
ai aws bedrock claude cloudwatch genkit golang llama nova plugin
Last synced: 21 days ago
JSON representation
AWS plugins for Google's GenKit framework - add AWS Bedrock models and CloudWatch monitoring to your existing GenKit applications
- Host: GitHub
- URL: https://github.com/scttfrdmn/genkit-aws
- Owner: scttfrdmn
- License: other
- Created: 2025-09-13T15:40:25.000Z (23 days ago)
- Default Branch: main
- Last Pushed: 2025-09-13T16:52:32.000Z (23 days ago)
- Last Synced: 2025-09-13T17:53:42.829Z (23 days ago)
- Topics: ai, aws, bedrock, claude, cloudwatch, genkit, golang, llama, nova, plugin
- Language: Go
- Size: 10.7 MB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Changelog: CHANGELOG.md
- Contributing: CONTRIBUTING.md
- Funding: .github/FUNDING.yml
- License: LICENSE
- Security: .github/SECURITY.md
Awesome Lists containing this project
README
# GenKit AWS Plugins
[](https://goreportcard.com/report/github.com/scttfrdmn/genkit-aws)
[](https://pkg.go.dev/github.com/scttfrdmn/genkit-aws)
[](https://opensource.org/licenses/Apache-2.0)
[](./coverage.html)
[](https://github.com/scttfrdmn/genkit-aws/releases)
[](https://github.com/scttfrdmn/genkit-aws/issues)
**AWS plugins for Google's GenKit framework** - add AWS Bedrock models and CloudWatch monitoring to your existing GenKit applications.
> **Important**: This is NOT a fork or port of GenKit. These are plugins that extend Google's GenKit framework to work with AWS services.
## What This Provides
- **AWS Bedrock Integration**: Use 40+ AWS models (Claude, Nova, Llama) through GenKit's standard API
- **CloudWatch Monitoring**: Automatic flow metrics and observability for GenKit applications
- **Seamless Integration**: Keep using GenKit's APIs, just add AWS capabilities
## Quick Start
### 1. Install Google's GenKit Framework First
```bash
go get github.com/firebase/genkit/go@latest
```
### 2. Add AWS Plugins
```bash
go get github.com/scttfrdmn/genkit-aws@latest
```
### 3. Use AWS Models in GenKit
```go
import (
// Google's GenKit framework
"github.com/firebase/genkit/go/genkit"
"github.com/firebase/genkit/go/ai"
// AWS plugins that extend GenKit
genkitaws "github.com/scttfrdmn/genkit-aws/pkg/genkit-aws"
"github.com/scttfrdmn/genkit-aws/pkg/bedrock"
"github.com/scttfrdmn/genkit-aws/pkg/monitoring"
)
func main() {
ctx := context.Background()
// Initialize GenKit with AWS plugins
err := genkit.Init(ctx, &genkit.Config{
Plugins: []genkit.Plugin{
genkitaws.New(&genkitaws.Config{
Region: "us-east-1",
Bedrock: &bedrock.Config{
Models: []string{"anthropic.claude-3-sonnet-20240229-v1:0"},
},
CloudWatch: &monitoring.Config{
Namespace: "MyApp/GenKit",
},
}),
},
})
if err != nil {
log.Fatal(err)
}
// Use GenKit's standard API with AWS models
resp, err := genkit.Generate(ctx, &ai.GenerateRequest{
Model: genkit.Model("anthropic.claude-3-sonnet-20240229-v1:0"), // Now available!
Messages: []*ai.Message{
{Role: ai.RoleUser, Content: []*ai.Part{{Text: "Hello!"}}},
},
})
}
```
## How It Works
1. **You use Google's GenKit framework** as normal
2. **Our plugins register AWS models** with GenKit's model registry
3. **GenKit's APIs work unchanged** - they just have more models available
4. **Monitoring automatically instruments** your existing GenKit flows
## Architecture
```
Your GenKit App
↓
Google's GenKit Framework (genkit.Model(), genkit.Generate(), etc.)
↓
Our AWS Plugins (genkitaws.New())
↓
AWS Services (Bedrock, CloudWatch)
```
## Available AWS Models
### Anthropic Claude
- `anthropic.claude-3-5-sonnet-20241022-v2:0`
- `anthropic.claude-3-sonnet-20240229-v1:0`
- `anthropic.claude-3-haiku-20240307-v1:0`
### Amazon Nova (Latest)
- `amazon.nova-pro-v1:0`
- `amazon.nova-lite-v1:0`
- `amazon.nova-micro-v1:0`
### Meta Llama
- `meta.llama3-2-90b-instruct-v1:0`
- `meta.llama3-2-11b-instruct-v1:0`
[See all 40+ supported models](docs/bedrock-models.md)
## Monitoring Integration
The CloudWatch plugin automatically tracks:
- **Flow Performance**: Duration, success rates, error types
- **Model Usage**: Token consumption, generation latency
- **Custom Metrics**: Configurable dimensions and namespaces
View in AWS CloudWatch under your configured namespace.
## Examples
### Basic Usage
```go
// Initialize GenKit + AWS plugins
err := genkit.Init(ctx, &genkit.Config{
Plugins: []genkit.Plugin{
genkitaws.New(&genkitaws.Config{
Region: "us-east-1",
Bedrock: &bedrock.Config{
Models: []string{"anthropic.claude-3-sonnet-20240229-v1:0"},
},
}),
},
})
// Use like any GenKit model
resp, _ := genkit.Generate(ctx, &ai.GenerateRequest{
Model: genkit.Model("anthropic.claude-3-sonnet-20240229-v1:0"),
Messages: []*ai.Message{{Role: ai.RoleUser, Content: []*ai.Part{{Text: "Hello"}}}},
})
```
### Multiple Providers
```go
// Mix Google AI, OpenAI, and AWS models in the same app
err := genkit.Init(ctx, &genkit.Config{
Plugins: []genkit.Plugin{
googleai.New(&googleai.Config{}), // Google's plugin
openai.New(&openai.Config{}), // OpenAI plugin
genkitaws.New(&genkitaws.Config{ // Our AWS plugin
Region: "us-east-1",
Bedrock: &bedrock.Config{
Models: []string{"anthropic.claude-3-sonnet-20240229-v1:0"},
},
}),
},
})
// All available through genkit.Model()
models := []string{
"googleai/gemini-1.5-pro", // Google
"openai/gpt-4", // OpenAI
"anthropic.claude-3-sonnet-20240229-v1:0", // AWS
}
```
### Flow with Monitoring
```go
// Define flow using GenKit's API
flow := genkit.DefineFlow("myFlow", func(ctx context.Context, input string) (string, error) {
resp, err := genkit.Generate(ctx, &ai.GenerateRequest{
Model: genkit.Model("anthropic.claude-3-sonnet-20240229-v1:0"),
Messages: []*ai.Message{{Role: ai.RoleUser, Content: []*ai.Part{{Text: input}}}},
})
if err != nil {
return "", err
}
return resp.Candidates[0].Message.Content[0].Text, nil
})
// Monitoring plugin automatically tracks this flow's performance
result, _ := flow.Run(ctx, "What's the weather?")
```
## Prerequisites
- Go 1.23+
- AWS credentials configured (`aws configure` or environment variables)
- AWS Bedrock model access enabled in your region
- Google's GenKit Go SDK
## Installation & Setup
### 1. Install Dependencies
```bash
go mod init my-genkit-app
go get github.com/firebase/genkit/go@latest
go get github.com/scttfrdmn/genkit-aws@latest
```
### 2. Configure AWS Credentials
```bash
aws configure
# or set environment variables:
# export AWS_REGION=us-east-1
# export AWS_ACCESS_KEY_ID=...
# export AWS_SECRET_ACCESS_KEY=...
```
### 3. Enable Bedrock Models
Go to AWS Console → Bedrock → Model Access and enable the models you want to use.
## Development
### Running Examples
```bash
cd cmd/examples/bedrock-chat
go run main.go "What is AI?"
cd cmd/examples/monitoring-demo
go run main.go
```
### Testing
```bash
make test # Unit tests
make test-integration # Integration tests (requires AWS credentials)
make lint # Linting
# Integration testing with custom AWS profile/region
make test-integration-custom AWS_PROFILE=myprofile AWS_REGION=us-east-1
# Standalone integration test script
./scripts/integration-test.sh aws us-west-2 # Custom profile and region
./scripts/integration-test.sh # Use defaults (aws profile, us-west-2)
```
### Integration Testing
The project includes comprehensive integration tests that verify real AWS Bedrock and CloudWatch functionality:
#### **Quick Integration Test**
```bash
# Test with your default AWS profile and us-west-2 region
./scripts/integration-test.sh aws us-west-2
```
#### **Prerequisites for Integration Tests**
- AWS CLI configured with valid credentials
- AWS Bedrock model access enabled in your target region
- Sufficient AWS permissions for Bedrock and CloudWatch
#### **What Integration Tests Verify**
- ✅ AWS authentication and configuration
- ✅ Bedrock model availability and access
- ✅ Model generation with real Claude/Nova models
- ✅ CloudWatch metrics submission and buffering
- ✅ Error handling with invalid models
- ✅ Flow monitoring and instrumentation
#### **Integration Test Options**
```bash
# Using Go test directly
AWS_PROFILE=aws AWS_REGION=us-west-2 go test -tags=integration ./test/integration/
# Using Makefile
make test-integration # Uses aws profile, us-west-2
make test-integration-custom AWS_PROFILE=prod AWS_REGION=us-east-1
# Using standalone script
./scripts/integration-test.sh aws us-west-2 10m # Custom timeout
```
**⚠️ Note**: Integration tests will incur small AWS charges for Bedrock API calls and CloudWatch metrics.
## Configuration
### Bedrock Configuration
```go
&bedrock.Config{
Models: []string{
"anthropic.claude-3-sonnet-20240229-v1:0",
"amazon.nova-pro-v1:0",
},
DefaultModelConfig: &bedrock.ModelConfig{
MaxTokens: 4096,
Temperature: 0.7,
TopP: 0.9,
StopSequences: []string{"STOP"},
},
ModelConfigs: map[string]*bedrock.ModelConfig{
"anthropic.claude-3-sonnet-20240229-v1:0": {
MaxTokens: 8192,
Temperature: 0.5,
},
},
}
```
### CloudWatch Configuration
```go
&monitoring.Config{
Namespace: "MyApp/GenKit",
EnableFlowMetrics: true,
EnableModelMetrics: true,
CustomDimensions: map[string]string{
"Environment": "Production",
"Version": "1.0.0",
},
MetricBufferSize: 20,
}
```
## Relationship to GenKit
| Component | Provided By |
|-----------|-------------|
| Core Framework | Google's GenKit |
| Flow API | Google's GenKit |
| Model Interface | Google's GenKit |
| Prompt Management | Google's GenKit |
| **AWS Bedrock Models** | **Our Plugin** |
| **CloudWatch Monitoring** | **Our Plugin** |
## Contributing
We welcome contributions! This project follows Google's GenKit plugin patterns and conventions.
### Development Setup
```bash
git clone https://github.com/scttfrdmn/genkit-aws
cd genkit-aws
go mod download
make test
```
### Submitting Issues
Please use GitHub Issues for bug reports and feature requests.
## License
Apache License 2.0 - see [LICENSE](LICENSE) for details.
---
**Note**: For GenKit framework documentation, see [Google's GenKit docs](https://firebase.google.com/docs/genkit). This repository only documents the AWS-specific plugins.