AI-assisted serverless development with AWS Lambda, SAM, CDK, API Gateway, EventBridge, and Step Functions.
This plugin provides comprehensive serverless development guidance with AWS best practices built-in. It integrates with the AWS Serverless MCP Server to give you access to the complete serverless development lifecycle.
Built as a Claude Code plugin, but the skill content follows the Agent Skills open standard and works with any compatible tool — Cursor, VS Code, Gemini CLI, OpenAI Codex, Roo Code, Amp, and many more. See Using with Other Tools below.
- Project Initialization: Create new SAM or CDK projects with templates and best practices
- Local Development: Test Lambda functions locally with Docker containers
- Build and Deploy: Compile, package, and deploy to AWS with CloudFormation
- Monitoring: Retrieve logs and metrics for debugging and optimization
- Full-Stack Apps: Deploy complete applications with Lambda Web Adapter
- Response Streaming: Stream responses from Lambda for LLM output, large payloads, and long-running operations
- Frontend Assets: Manage S3 hosting with CloudFront distribution
- Custom Domains: Configure Route 53 DNS and ACM certificates
- Event Source Mappings: Configure Lambda triggers for S3, SNS, DynamoDB, Kinesis, SQS, Kafka, MQ, DocumentDB
- EventBridge Integration: Event bus design, routing rules, Pipes, archive and replay
- Schema Registry: Type-safe event handling with schema discovery
- Lambda Durable Functions: Multi-step workflows with checkpointing for Python 3.14+ and Node.js 22+
- Step Functions: Visual orchestration with 220+ AWS service integrations
- Patterns: Sequential, parallel, human-in-the-loop, and saga patterns
- Structured Logging: JSON logging with Powertools Logger, log level strategy, decorator stacking order
- Distributed Tracing: X-Ray with Powertools Tracer, annotations, subsegments, cold start filtering
- CloudWatch Application Signals: APM with ADOT, service maps, SLO tracking, anomaly detection
- Custom Metrics: Embedded Metric Format (EMF), business KPI vs technical metrics
- Dashboards and Alarms: Two-tier dashboard strategy, p90/p99 latency alarms, composite alarms
- Performance Tuning: Memory right-sizing, Lambda Power Tuning, cold start optimization, SnapStart, Managed Instances
- Cost Optimization: Right-sizing, reserved concurrency, architecture recommendations
- Troubleshooting: Symptom-based diagnosis for common serverless issues
- SAM Template Validation: Automatically runs
sam validateafter Claude editstemplate.yamlortemplate.yml, catching syntax and configuration errors immediately
- AWS CLI configured with credentials
- AWS SAM CLI installed
- Docker Desktop (for local testing)
- Python 3.10+ with uv package manager
This plugin's MCP server is configured with two flags that grant elevated access:
--allow-write: Enables write operations such as creating SAM projects, deploying stacks, and modifying AWS resources. Without this flag, the server operates in read-only mode.--allow-sensitive-data-access: Enables access to Lambda function logs and API Gateway logs, which may contain sensitive application data.
To restrict these capabilities, edit .mcp.json and remove the corresponding flags from the args array.
The plugin includes a hook that runs sam validate after Claude edits SAM template files. This requires SAM CLI to be installed. If SAM CLI is not found, the hook silently skips validation.
To disable the hook, run /hooks in Claude Code and toggle it off. If the hook doesn't run after cloning, ensure the script is executable: chmod +x scripts/validate-template.sh.
-
Add the marketplace in Claude Code:
/plugin marketplace add gunnargrosch/gunnargrosch-plugins
-
Install the plugin:
/plugin install aws-serverless@gunnargrosch-plugins
This plugin bundles two things:
- A skill (
/aws-serverless): Workflow guidance, best practices, and troubleshooting references that teach Claude how to build serverless applications effectively - An MCP server (AWS Serverless MCP Server): Tools for the full serverless lifecycle — project initialization, builds, deployments, testing, observability, and security policy generation
Note: The MCP server is configured to use
@latestso you always get the newest tools and bug fixes. The package is pre-1.0 and under active development. To pin to a specific version for reproducibility, see the "MCP Server Configuration" section in SKILL.md.
The plugin automatically activates when you mention serverless-related topics including:
serverless,lambda,sam,cdkapi gateway,function url,lambda web adapterdynamodb,kinesis,sqs,kafkaevent-driven,event source mapping,cold starteventbridge,event bus,step functions,durable functions,state machine
You can also invoke the skill directly:
/aws-serverlessCreate a new serverless project:
I want to build a serverless API for a todo application using Python and AWS Lambda. Can you help me set up a new SAM project?
Deploy a web application:
I have a React frontend and Express.js backend. Help me deploy this as a full-stack serverless application on AWS.
Set up event processing:
I need to process messages from an SQS queue with Lambda. Help me configure the event source mapping with batch processing and error handling.
Design an event-driven architecture:
I'm building an order processing system. Help me set up EventBridge with custom events and multiple consumer services.
aws-serverless-plugin/
├── .claude-plugin/
│ └── plugin.json # Plugin manifest
├── .mcp.json # MCP server configuration
├── skills/
│ └── aws-serverless/
│ ├── SKILL.md # Main skill definition
│ ├── getting-started.md # Decision tree: what are you building?
│ ├── sam-project-setup.md # SAM project initialization and workflow
│ ├── cdk-project-setup.md # CDK constructs, testing, pipelines, SAM coexistence
│ ├── web-app-deployment.md # Full-stack deployment with Lambda Web Adapter
│ ├── event-sources.md # Lambda event sources: S3, SNS, DynamoDB, Kinesis, SQS, Kafka, MQ, DocumentDB
│ ├── event-driven-architecture.md # EventBridge, event design, Pipes, schema registry
│ ├── orchestration-and-workflows.md # Durable Functions, Step Functions, workflow patterns
│ ├── optimization.md # Performance, cost, streaming, Powertools
│ ├── observability.md # Logging, tracing, metrics, dashboards, Application Signals
│ └── troubleshooting.md # Symptom-based diagnosis and resolution
├── hooks/
│ └── hooks.json # Hook configuration (SAM template validation)
├── scripts/
│ └── validate-template.sh # Runs sam validate after template edits
├── CHANGELOG.md # Version history
├── README.md # This file
└── LICENSE # MIT License
The skill content in skills/aws-serverless/ follows the Agent Skills open standard — a portable format for giving AI agents new capabilities, supported by Claude Code, Cursor, VS Code, Gemini CLI, OpenAI Codex, Roo Code, Amp, and many more.
Any tool that supports Agent Skills can use the skills/aws-serverless/ directory directly. For tools without native support, the content is plain markdown that works as context files or system prompt input.
git clone https://github.com/gunnargrosch/aws-serverless-plugin.gitCursor — copy into .cursor/rules/:
cp -r skills/aws-serverless/ your-project/.cursor/rules/aws-serverless/VS Code / Copilot — copy into .github/instructions/:
cp -r skills/aws-serverless/ your-project/.github/instructions/aws-serverless/Gemini CLI — copy into .gemini/instructions/:
cp -r skills/aws-serverless/ your-project/.gemini/instructions/aws-serverless/Note: The MCP server and hooks are Claude Code specific. Other tools get the skill content (guides, best practices, troubleshooting) but not the MCP server integration or SAM template validation hook.
This plugin is open source and contributions are welcome! Please feel free to:
- Report issues or suggest improvements
- Submit pull requests with enhancements
- Share your serverless patterns and best practices
- Help improve the documentation
This project is licensed under the MIT License - see the LICENSE file for details.