corgivable

Terraform IaC for a serverless Discord + Instagram bot stack on AWS Lambda.

Personal portfolio / learning project. A small but complete reference for wiring AWS Lambda, API Gateway, SNS, SQS, S3, DynamoDB, EventBridge and SSM Parameter Store together with Terraform Cloud-managed state.

What it does

A pipeline that drafts AI image prompts in Discord, renders them externally, stores the resulting images in S3, and publishes them to Instagram on a scheduled cron. End-to-end serverless, single region (eu-central-1).

Architecture

   Discord slash cmd
          |
          v
   API Gateway (HTTP)
          |
          v
   Lambda: pull-save-image  ---->  SQS: inbound-prompt-drafts
                                          |
                                          v
                                   (external renderer
                                    picks up message)
                                          |
                                          v
                                    SNS: ready-prompt
                                          |
                                          v
                              Lambda: discord-prompt-poster
                                          |
                                          v
                                  S3: corgi-rendered-images-ready
                                  DynamoDB: corgi-meta-data
                                          |
   EventBridge cron (daily 16:00 UTC) --> Lambda: publish-image  --> Instagram
                                          ^
                                          |
                                  SSM Parameter Store
                                  (IG creds, SecureString)

Components

Component	Purpose
discord_pull_image Lambda	Receives Discord interaction, enqueues prompt draft to SQS
discord_prompt_poster Lambda	Subscribed to SNS, posts the ready prompt back into Discord
publish_image Lambda	Cron-triggered; reads from S3 + DynamoDB, publishes to Instagram
extract_images Lambda	Helper for pulling images out of the bucket (gated behind API Gateway, currently disabled in the route map)
API Gateway v2 (HTTP)	Public entry point for Discord interactions
SQS inbound-prompt-drafts	Buffers prompts awaiting render
SNS ready-prompt	Fan-out for rendered prompts
S3 corgi-rendered-images-ready	Golden source of generated images (intentionally force_destroy = false)
DynamoDB corgi-meta-data	Tracks which images have been posted
SSM Parameter Store	SecureString storage for Instagram credentials
Lambda Layer node_modules	Shared node_modules/ for all Node.js Lambdas

Tech

• IaC: Terraform, state managed in Terraform Cloud (organization = "basil-home", workspace = "corgifun")
• Cloud: AWS, region eu-central-1
• Runtime: Node.js 16.x (Lambda)
• Helper: a Python prompt generator (see python-prompt-generator/)
• CI: GitHub Actions workflow in .github/workflows/terraform.yml

Secrets and variables

All sensitive inputs are declared in variables.tf with sensitive = true and supplied via the Terraform Cloud workspace (never committed):

Variable	Description
discord_token	Discord bot token
discord_public_key	Discord application public key (interaction verification)
ig_username	Instagram account username
ig_password	Instagram account password

Instagram credentials are additionally written to SSM Parameter Store as SecureString so the publishing Lambda can read them at runtime without embedding them in environment variables.

Note on history: an early CI workflow embedded AWS credentials directly. Those credentials were rotated long ago and the history of this repo has been rewritten to remove the literal values before this README was added.

Repository layout

.
├── 00-infra-vpc-main.tf            # (currently inert) VPC scaffolding
├── 00-lambda-nodemodules-layer.tf  # Shared Lambda layer
├── 01-*.tf                         # Discord-side resources (Lambda, SQS, SNS, API GW)
├── 02-*.tf                         # Image processing Lambda + S3/DynamoDB/API GW
├── 03-publish-image-to-ig-lambda.tf# IG publisher Lambda + EventBridge cron + SSM
├── 04-extract-images-lambda.tf     # Image-extraction Lambda
├── lambda/                         # Node.js Lambda sources
├── lambda-layer/                   # node_modules bundled into a Lambda layer
├── python-prompt-generator/        # Auxiliary prompt-generation script
├── register-discord-commands/      # One-shot helper for registering slash commands
├── variables.tf                    # Sensitive input declarations
├── provider.tf                     # AWS provider + Terraform Cloud backend
└── outputs.tf                      # Emits the API Gateway endpoint URL

Deploy

Prerequisites: Terraform >= 1.x, an AWS account, a Terraform Cloud workspace called corgifun under organization basil-home (or fork and adjust provider.tf).

# 1. Authenticate against Terraform Cloud
terraform login

# 2. Initialize providers + remote state
terraform init

# 3. Set the four sensitive variables in the TFC workspace UI
#    (discord_token, discord_public_key, ig_username, ig_password)

# 4. Plan / apply
terraform plan
terraform apply

The api_gateway_url output is the endpoint to register with Discord as the interactions URL.

Destroy

terraform destroy

Note that the corgi-rendered-images-ready S3 bucket has force_destroy = false — generated images are treated as the golden source and must be emptied manually before the bucket can be destroyed.

Status

Portfolio / learning project. Not maintained for production use. Useful as a small but complete reference for a Lambda + API Gateway + SNS/SQS pipeline managed entirely from Terraform with remote state.

License

No license is granted; all rights reserved by the author. Open an issue if you want to reuse a piece of it.