platform-architect

Files (2)

• EXAMPLE.md
• SKILL.md

SKILL.md (first lines)

# Platform Architect Skill

## Overview

Design and architect cloud infrastructure that is secure, production-ready, and
compliant with organization standards. 

Output the plan for deploying and implementing infrastructure in a proposal document called `ARCHITECTURE.md`.

---

# Process

## Phase 1: Prerequisites

Get the following information from the user:

- HCP Terraform organization: default `alice-publishing`
- HCP Terraform project: default `applications`
- Environment: default `nonprod`, can be `nonprod` or `prod`
- OwnedBy
- BusinessUnit
- Project
- Purpose

Before proceeding to Phase 2, verify:
- [ ] User has specified all information for HCP Terraform
- [ ] User has specified all information for tags

## Phase 2: Design

### Identify required infrastructure resources

- Search architecture documentation for the use case
  - [AWS Architecture Center](https://aws.amazon.com/architecture/reference-architecture-diagrams)
  - [HashiCorp Well-Architected Framework](https://developer.hashicorp.com/well-architected-framework)
- If architecture guidance does not exist, conduct a broader web search.
- Identify service offerings required to achieve architecture.
- Generate a list of products and services to use to achieve the architecture.

If the user specifies that they must use an existing resource (such as a VPC or Kubernetes cluster),
get more information on the tags related to the existing resource.

Ask for clarification if you find missing information.

AWS Requirements:

- Create networking (e.g., VPC) or runtime (e.g., EC2 instances) unless otherwise
  specified by the user
- Store secrets in AWS Secrets Manager.
- Create IAM roles and security groups for components.
- Use AWS ALB for load balancing services, when possible

### Discover available Terraform modules and Helm charts

1. Check if private Terraform modules or Helm charts exist for a resource
2. Only use raw resources if:
   - No private module or public Helm chart exists
   - The module or chart doesn't meet specific requirements
3. Document why raw resources are used instead

#### Helm

Search for Helm charts using `helm search hub <keyword>`.

Document in table format:
```markdown
| Service | Helm Chart Available? | Repository | Chart Name | Chart Version |
|---------|----------------------|------------|------------|---------|
| Qdrant  | Yes                  | https://qdrant.github.io/qdrant-helm | qdrant | 0.8.4 |
```

#### Terraform

Search for private modules with `search_private_modules` (organization-specific only).
Check latest stable version with `get_private_module_details`.

Document in table format:
```markdown
| Resources | Module Source | Version |
|---------|----------------------|------------|
| Cloudflare DNS Record | app.terraform.io/alice-publishing/alb-dns/cloudflare | 0.0.1 |
```

### Other Kubernetes resources

For other Kubernetes resources not addressed by Helm charts:

- Use AWS ALB Ingress
- Create a service account on Kubernetes for each service
- Each service has its own namespace, do not deploy to default

List required resources: Deployment, PVC, Service, ServiceAccount, Ingress, Secret

### Other Terraform resources

For other Terraform resources not addressed by private modules:

- `get_latest_provider_version` to get provider version
- `get_provider_details` to evaluate if the provider offers the resources required.
- Using the list of products and services to achieve the architecture,
find their equivalent resources and data sources in the provider
with `get_provider_capabilities`.
- If the user requests the resources be created on an existing resource,
suggest a list of Terraform data sources to use.

Document in sections:
- Providers (name and version)
- Data Sources (list)
- Resources (organized by category: Networking, Compute, Storage, etc.)

### Checklist

- [ ] User has approved the resource list
- [ ] All required variables have been identified
- [ ] Architecture aligns with AWS/HashiCorp best practices
- [ ] Private modules identified for all applicable resources
- [ ] Helm charts identified for Kubernetes services
- [ ] User understands the deployment approach

From joatmon08/platform-infrastructure-skills, MIT.

More from joatmon08/platform-infrastructure-skills