Documentation

Calor documentation - Learn how to use the Coding Agent Language for Optimized Reasoning

Welcome to the Calor documentation. Calor is a programming language designed specifically for AI coding agents, compiling to .NET via C# emission.

Quick Links

Getting Started

New to Calor? Start here to learn the basics:

• Installation - Set up the Calor compiler
• Hello World - Write your first Calor program
• Claude Integration - Use Calor with AI agents

Syntax Reference

Complete language reference:

• Structure Tags - Modules, functions, block structure
• Types - Type system, Option, Result
• Expressions - Lisp-style operators
• Control Flow - Loops, conditionals
• Contracts - Requires, ensures
• Effects - Effect declarations

CLI Reference

Command-line tools:

• calor - Compile Calor to C#
• calor assess - Score C# files for migration potential
• calor import - Generate honest effect coverage from a package
• calor review-packet - Review the unproven remainder and waivers
• calor verify - Interpret seven-status contract results

v0.12 Adoption and Trust

• Adoption Playbook - Add Calor one module at a time and keep an exit path
• Verification Guarantees - Modeled semantics, assumptions, and runtime checks
• Telemetry and Privacy - Opt-in behavior and exact metadata inventory

Benchmarking

Evaluation framework and results:

• Results - Benchmark comparison with C#
• Methodology - How benchmarks are measured

Formal Semantics

Backend-independent language specification:

• Overview - Why formal semantics matter for agents
• Core Semantics - Evaluation order, scoping, numerics, contracts
• Normal Form (CNF) - Intermediate representation
• .NET Backend - How the backend implements semantics
• Versioning - Semantic versioning for stability

Philosophy

Design decisions and rationale:

• Why Calor Exists - The motivation behind Calor
• Design Principles - Core design principles
• Tradeoffs - What Calor gives up for explicitness

Contributing

Get involved:

• Development Setup - Set up your dev environment
• Adding Benchmarks - Contribute evaluation programs