At a Glance:
Encore is an infrastructure-from-code orchestration platform that derives cloud resources directly from application code, runs the same backend on local, AWS, and GCP environments without writing Terraform or YAML, and deploys to your own cloud account.
Overview:
Encore is a backend development platform that derives infrastructure from application code and manages it consistently across local and cloud environments. Developers declare resources such as SQL databases, Pub/Sub, object storage, caches, cron jobs, and secrets semantically inside their application code. Encore builds an application graph from these declarations and provisions the matching infrastructure for each environment: local sandbox environments, managed cloud resources on AWS or GCP, or self-hosted Docker images. It replaces separate Infrastructure-as-Code tooling with a workflow where infrastructure stays in sync with the code that uses it. Encore targets teams building type-safe, traceable backends in Go or TypeScript who want to deploy into their own AWS or GCP accounts without managing Terraform configurations.
Key Decision Points:
Infrastructure is derived from code, not written separately: Encore eliminates the need for separate IaC by generating infrastructure provisioning from resource declarations embedded in application code.
Deploys to your own cloud account: Encore provisions resources directly into your existing AWS or GCP account rather than running on a managed third-party runtime.
Language constraint: Services must be written in Go or TypeScript (Node.js), with Python support listed as coming soon.
Infrastructure scope is focused on common resources: Encore handles SQL databases, Pub/Sub, object storage, caches, cron jobs, and secrets, but does not prevent integrating any other infrastructure via SDKs or the Terraform provider.
You can adopt incrementally: New Encore services can run alongside an existing stack, and migration can proceed service by service.
Core Features:
Application graph: Encore parses resource declarations from application code and builds a graph that drives both local development environments and cloud provisioning.
Environment-aware provisioning: The same resource declaration provisions local equivalents during development (e.g., Postgres locally, NSQ for Pub/Sub) and managed cloud services on deploy (e.g., RDS, SNS+SQS, Cloud SQL, Cloud Pub/Sub).
Semantic resource declarations: Developers declare resources in code using semantics like “I need a Postgres database” rather than specifying infrastructure configuration directly.
Preview environments: Each pull request receives a cloud preview environment for end-to-end testing before merging.
Built-in local development dashboard: The CLI serves a dashboard at localhost:9400 for inspecting services, APIs, traces, databases, Pub/Sub messages, and architecture diagrams.
AI-assisted development support: Every Encore app includes a CLAUDE.md file and an MCP server that agents can use to introspect the application and generate type-safe code.
Use Cases:
Developer teams building Go or TypeScript backends that need to run identically on local machines and in AWS or GCP without maintaining separate IaC configurations.
Teams migrating from manually configured local development environments (Localstack, docker compose) to a workflow where infrastructure is derived from application code.
Use cases where cloud preview environments for pull requests are needed for end-to-end testing before deployment.
Open-Source Alternative Value:
Encore provides an open-source approach to infrastructure provisioning by deriving cloud resources from application code rather than requiring separate Infrastructure-as-Code tooling. Developers can run Encore entirely without its managed cloud platform by using encore build docker to produce standalone images that can be deployed with a custom infra configuration. The platform works within the user's own AWS or GCP account, and application code is written in standard Go or TypeScript, which the project states makes migration away from Encore straightforward since 99% of the code remains regular language code without framework lock-in.
