Local-to-AWS Parity
The whole point of the Floci emulator is parity: build and test against AWS APIs locally, then deploy to the real managed services with no surprises. Every marketplace recipe is tagged with the AWS service it targets, a parity note, and the one step to switch from local to production.
You can see this on each recipe card in the Marketplace (the “Deploys to …” badge), in each recipe’s README.md under Path to AWS, and in the machine-readable aws block of every recipe.json — which the MCP server exposes so your AI agent can guide the migration.
Two kinds of parity
Section titled “Two kinds of parity”1. AWS-SDK clients wired straight to Floci. A few recipes are AWS tools pointed at the emulator on 4566, so what you browse locally is the emulator’s state:
| Recipe | Wired to | In production |
|---|---|---|
dynamodb-admin | Floci DynamoDB (4566) | Amazon DynamoDB console |
s3-admin | Floci S3 (4566) | Amazon S3 console |
Switch them to production by repointing their endpoint variable (DYNAMO_ENDPOINT / S3_ENDPOINT) at a real regional endpoint — or removing it.
2. Engine-compatible stand-ins. Most recipes run the same engine AWS manages, so your app code, drivers and wire protocol are identical locally and in the cloud. Moving to production is a connection-string swap, not a rewrite.
Recipe → managed AWS service
Section titled “Recipe → managed AWS service”| Recipe | Managed AWS service | How you switch |
|---|---|---|
| PostgreSQL | Amazon RDS for PostgreSQL / Aurora | Repoint the connection string |
| Redis | Amazon ElastiCache for Redis | Point REDIS_URL at the cluster endpoint |
| MongoDB | Amazon DocumentDB | Repoint the connection string (+ RDS CA) |
| ClickHouse | Self-host on ECS/EKS (alt: Redshift) | Run the image on ECS, attach a volume |
| RabbitMQ | Amazon MQ for RabbitMQ | Swap the AMQP connection URI |
| NATS JetStream | Amazon SNS + SQS (or Amazon MQ) | Map subjects→SNS, consumers→SQS |
| Redpanda | Amazon MSK (Kafka) | Point bootstrap servers at MSK |
| MinIO | Amazon S3 | Drop the custom endpoint from the SDK |
| Meilisearch | Amazon OpenSearch Service | Migrate the index / self-host on ECS |
| Qdrant | OpenSearch Serverless (vector) / pgvector | Recreate the collection, repoint client |
| Keycloak | Amazon Cognito | Swap the OIDC issuer/JWKS URL |
| Vault | AWS Secrets Manager + KMS | Migrate secrets, swap the SDK |
| Mailpit | Amazon SES | Point SMTP at the SES relay |
| Ollama | Amazon Bedrock | Swap the base URL/SDK to Bedrock Runtime |
| Temporal | Self-host on EKS (alt: Step Functions) | Run on EKS with RDS persistence |
| n8n | Self-host on Amazon ECS/EKS | Run on ECS with EFS + RDS |
| Jaeger | AWS X-Ray (via ADOT) | Export OTLP to the ADOT collector |
| Grafana + Prometheus | Managed Grafana + AMP | Remote-write to AMP, import dashboards |
| Nginx Proxy Manager | AWS ALB + ACM | Recreate hosts as ALB rules |
| PocketBase | Self-host on ECS/Fargate | Run on Fargate with an EFS volume |
| Portainer | Amazon ECS / EKS console | Push to ECR, define ECS/EKS workloads |
| IoT Core (MQTT) | AWS IoT Core | Repoint MQTT clients at the ATS endpoint |
| Transfer Family (SFTP) | AWS Transfer Family | Provision a Transfer Family SFTP server |
| DynamoDB Admin | Amazon DynamoDB | Repoint DYNAMO_ENDPOINT at the region |
| S3 Admin | Amazon S3 | Repoint S3_ENDPOINT at the region |
💡 Pair this with the production paths in the README (AWS Copilot, Terraform/CDK, SAM) to ship the whole stack.