Pkl — Configuration as Code Language by Apple

What it is

Pkl is a programmable configuration language developed by Apple. It provides typed schemas, validation constraints, templating, and IDE support for configuration files. Instead of writing raw YAML or JSON with no validation until runtime, you define configuration as Pkl code with types, defaults, and constraints that are checked at evaluation time.

Pkl is designed for platform engineers, DevOps teams, and application developers who manage complex configuration across multiple environments and want compile-time safety.

How it saves time or tokens

YAML and JSON configuration files are error-prone: a mistyped key, a wrong type, or a missing field causes runtime failures. Pkl catches these errors at evaluation time before deployment. The language supports modules, imports, and inheritance, so you can define a base configuration once and override per environment. This eliminates copy-paste duplication across dev/staging/production configs and reduces configuration-related incidents.

How to use

1. Install the Pkl CLI:

curl -L https://github.com/apple/pkl/releases/latest/download/pkl-linux-amd64 -o pkl
chmod +x pkl && sudo mv pkl /usr/local/bin/

1. Create a Pkl configuration file:

// config.pkl
module Config

name: String = 'my-app'
port: Int(this >= 1024 && this <= 65535) = 8080
debug: Boolean = false
database {
  host: String = 'localhost'
  port: Int = 5432
  name: String = 'mydb'
}

1. Evaluate to YAML or JSON:

pkl eval config.pkl -f yaml
pkl eval config.pkl -f json

Example

Environment-specific configuration with inheritance:

// base.pkl
module BaseConfig

name: String
replicas: Int = 1
resources {
  cpu: String = '100m'
  memory: String = '256Mi'
}

// production.pkl
amends "base.pkl"

name = 'api-server'
replicas = 3
resources {
  cpu = '500m'
  memory = '1Gi'
}

pkl eval production.pkl -f yaml outputs the merged configuration with production overrides applied.

Related on TokRepo

• DevOps tools — Browse infrastructure and configuration tools
• Automation tools — Explore configuration management automation

Common pitfalls

• Trying to use Pkl as a general-purpose programming language. Pkl is designed for configuration, not application logic. It has deliberate limitations (no loops with side effects, no I/O) to keep configurations deterministic.
• Not leveraging the type system. Writing Pkl without type annotations and constraints misses the main benefit. Add types and validation constraints to catch errors before deployment.
• Forgetting that Pkl evaluates to a data format (YAML, JSON, plist). Your tooling consumes the output, not Pkl files directly. Ensure your CI pipeline includes a pkl eval step.