calor

Calor is a programming language designed specifically for AI coding agents, compiling to .NET via C# emission. Traditional languages are built for humans, forcing AI agents to infer semantic meaning—such as side effects, contracts, and scope boundaries—from implicit syntax. Calor eliminates this mismatch by making these elements explicit in the language itself. Core principles include explicit effect declarations (e.g., §E{cw, fs:r, net:rw}), first-class preconditions and postconditions (§Q and §S), optional stable identifiers for precise referencing across refactors, indent-based blocks to reduce token cost, and Lisp-style operators for machine-readable semantics. This design trades raw token efficiency for semantic clarity, enabling agents to reason about code behavior, detect contract violations, and perform precise edits without deep parsing. Calor ships with a first-class language server (LSP) supporting diagnostics, go-to-definition, references, rename, formatting, and semantic tokens. It also includes an MCP (Model Context Protocol) server exposing 19 tools for compilation, verification, code navigation, analysis, migration, linting, and syntax help, allowing AI agents to interact directly with the compiler. Quick-start integration is provided for Claude Code, OpenAI Codex CLI, Google Gemini CLI, and GitHub Copilot, with auto-configuration of project instructions and MCP settings. The project is open to contributions, especially in benchmarking, parser improvements, and documentation.