arig

A polyglot service orchestrator. Reads a YAML file describing a set of services with their commands and dependencies, then builds and runs them locally with the right startup ordering.

Early development. Only process supervision and dependency ordering work today. The eventual goal is one tool that runs services from source for local dev, and emits Kubernetes/Helm manifests for the same config.

Usage

arig up

Reads arig.yaml in the current directory. Example:

services:
  db:
    command: docker run --rm -p 5432:5432 -e POSTGRES_PASSWORD=dev postgres:16

  migrate:
    command: ./scripts/migrate.sh
    type: oneshot
    depends_on: [db]

  api:
    command: cargo run
    working_dir: ./api
    depends_on: [migrate]

Editor integration

A JSON schema for arig.yaml is checked in at arig.schema.json. With the YAML language server installed, add a directive at the top of your config:

# yaml-language-server: $schema=https://raw.githubusercontent.com/cienvita/arig/main/arig.schema.json

To match the schema to your installed binary instead, generate it locally:

arig schema > arig.schema.json
# yaml-language-server: $schema=./arig.schema.json

Todo

Near-term:

• -C dir flag (chdir before reading config)
• Resolve working_dir and template paths against the yaml file's directory
• HTTP/TCP health checks with readiness gating
• Template rendering for .arig/templates -> .arig/generated
• Dynamic env injection from dependency metadata

Command surface:

• Implement arig down
• --format json structured output
• Single-service commands (status, logs, env, restart, build)

Plugin platform:

• Docker runtime via bollard
• Helm/k8s publish plugins
• External plugin protocol
• arig mcp server

License

Dual-licensed under either of MIT or Apache-2.0, at your option.