How aregm/icl works
ICL is a Python SDK for describing infrastructure, runtime dependencies, and programs through a unified API, then running workloads locally or on cloud infrastructure. The principal workflow is developer → define infrastructure/runtime/program → deploy and run → observe results and logs. Additional capabilities include Prefect flows, Docker image builds, cloud/infrastructure plugins, and JupyterHub-based development environments. The graph emphasizes relationships visible in excerpts or explicitly documented in the README; unsampled API and plugin behavior is represented conservatively.
Main entry points: remote.py Cluster Docker · runtime.py Prefect Runtime · prefect_runtime/program.py Prefect Program · api/program.py Program API
Components 27
Developer API
- Infrastructure API
src/infractl/api/infrastructure.py - Program API
src/infractl/api/program.py - Runtime API
src/infractl/api/runtime.py - Deploy API
src/infractl/api/deploy.py - Run API
src/infractl/api/run.py
Execution Core
- Infrastructure Model
src/infractl/base/infrastructure.py - Program Model
src/infractl/base/program.py - Runtime Model
src/infractl/base/runtime.py - Plugin Registry
src/infractl/base/registry.py - Identity Utilities
src/infractl/identity.py - Logging
src/infractl/logging.py
Runtime Adapters
- Image Builder
src/infractl/docker/builder.py - Local Docker
src/infractl/docker/local.py - Cluster Docker
src/infractl/docker/remote.py - Prefect Runtime
src/infractl/plugins/prefect_runtime/runtime.py - Prefect Program
src/infractl/plugins/prefect_runtime/program.py - Kubernetes Runtime
src/infractl/plugins/kubernetes_runtime/runtime.py - AWS Networking
src/infractl/plugins/aws_infrastructure/vpc.py - ICL Infrastructure
src/infractl/plugins/icl_infrastructure/infrastructure.py - Kubernetes API
src/infractl/kubernetes.py
Development Environments
- JupyterHub
src/infractl/hub/jupyterhub.py - Hub Configuration
src/infractl/hub/config.py - Hub Server
src/infractl/hub/server.py
Other components
- Developer
- Prefect Service
- Cloud Infrastructure
- Python Workload
Connections 22
- Developer → Infrastructure API: defines infrastructure
- Developer → Program API: defines program
- Developer → Run API: runs workload
- Program API → Program Model: creates
src/infractl/api/program.py - Infrastructure API → Infrastructure Model: resolves
src/infractl/api/infrastructure.py - Runtime API → Runtime Model: configures
- Deploy API → Prefect Runtime: deploys
- Run API → Python Workload: executes
- Infrastructure Model → Plugin Registry: resolves plugins
- Image Builder → Local Docker: selects builder
src/infractl/docker/builder.py - Image Builder → Cluster Docker: selects builder
src/infractl/docker/builder.py - Cluster Docker → Deploy API: deploys builder flow
src/infractl/docker/remote.py - Cluster Docker → Program API: creates builder program
src/infractl/docker/remote.py - Prefect Runtime → Prefect Program: deploys programs
src/infractl/plugins/prefect_runtime/runtime.py - Prefect Runtime → Prefect Service: uses orchestration
src/infractl/plugins/prefect_runtime/runtime.py - Prefect Program → Prefect Service: reads runs and logs
src/infractl/plugins/prefect_runtime/program.py - Kubernetes Runtime → Kubernetes API: uses Kubernetes
- ICL Infrastructure → Cloud Infrastructure: targets infrastructure
- AWS Networking → Cloud Infrastructure: provides networking
- JupyterHub → Hub Configuration: loads configuration
src/infractl/hub/jupyterhub.py - Hub Server → JupyterHub: serves hub
- Prefect Program → Identity Utilities: sanitizes names
src/infractl/plugins/prefect_runtime/program.py