---

name: serverless description: > Advanced serverless deployment and cold-start optimization skill. version: "2.0.0" author: "j4flmao" license: "MIT" type: skill compatibility: claude-code: true cursor: true codex: true windsurf: true tags: [serverless, aws, lambda, terraform]

Serverless Mastery

Purpose

Comprehensive serverless architecture management, focusing on AWS Lambda cold-start mitigation, Terraform IaC provisioning, and highly scalable event-driven patterns.

Core Principles

1. Minimize deployment package size to reduce cold start latency.
2. Leverage provisioned concurrency for latency-sensitive endpoints.
3. Optimize execution environments using Graviton and minimal runtimes.
4. Apply principle of least privilege in IAM roles.
5. Use Infrastructure as Code (Terraform) for reproducible environments.

Agent Protocol

• Triggers: "deploy lambda", "optimize cold start"
• Input Context Required: AWS credentials, source code, traffic patterns
• Output Artifact: Terraform plan, optimized deployment zip
• Response Formats:

  {
    "status": "deployed",
    "lambda_arn": "arn:aws:lambda:us-east-1:123456789012:function:my-func",
    "cold_start_est_ms": 120
  }
  

Decision Matrix

Is workload latency-sensitive?
 +-- Yes --> Use Provisioned Concurrency
 +-- No --> Standard on-demand Lambda

Detailed Architectural Overview

[API Gateway] ---> [Lambda Function] ---> [DynamoDB]
                   (Graviton2/Python)

Workflow Steps

Phase 1: Analysis

1. Inspect source
2. Identify deps
3. Measure size
4. Check limits

Phase 2: Optimization

1. Strip binaries
2. Bundle code
3. Set architecture
4. Tune memory

Phase 3: IaC Generation

1. Write main.tf
2. Define variables.tf
3. Setup outputs.tf
4. Provider config

Phase 4: Deployment

1. terraform init
2. terraform plan
3. terraform apply
4. verify state

Phase 5: Validation

1. Invoke lambda
2. Measure latency
3. Verify logs
4. End-to-end test

Phase 6: Monitoring

1. Setup CloudWatch
2. Create alarms
3. Dashboard config
4. Set alerts

Extended Troubleshooting Guide

Symptom	Primary Cause	Mitigation Action
Timeout	Cold start	Increase timeout or use provisioned concurrency
Missing Module	Bad bundling	Check deployment package contents
Access Denied	IAM role	Attach correct policies
Throttling	Concurrency limit	Request quota increase
High cost	Over-provisioned	Tune memory/concurrency
Missing logs	CW permissions	Add AWSLambdaBasicExecutionRole

Complete Execution Scenario

Start -> Build -> Optimize -> Plan -> Apply -> Done

Rules and Guidelines

1. Always bundle dependencies.
2. Never hardcode credentials.
3. Use Terraform state locking.
4. Tag all resources.
5. Implement dead-letter queues for async lambdas.

Reference Guides

• Architecture Guide
• Cold Start Opt
• Terraform Core
• Security Model
• Monitoring Setup
• Cost Tuning
• Advanced Networking
• CI/CD Pipelines

Handoff

Refer to aws-core skill for baseline AWS authentication.