ZeroLang Overview

ZeroLang is an experimental, graph-first programming language designed for agents rather than humans. It emphasizes semantic program structures over raw source text, enabling more precise interactions with code.

Key Features

• Graph-First Approach: Rather than focusing solely on source text, ZeroLang uses a graph structure for programming. This allows agents to make alterations by editing the program graph instead of simply modifying text.

• Source as Truth: While the graph is used for interactions, the original source text remains the durable, human-readable representation, facilitating code reviews and audits.

• Efficiency Goals: ZeroLang prioritizes token efficiency, low memory usage, fast startup and builds, low runtime latency, and zero dependencies.

Graph-Based Programming

• Why a Graph?: The graph serves as a better interface for program understanding, allowing agents to gather specific contextual information and make validated changes without relying on potentially stale text-based edits.

• Semantic Navigation: Agents can directly navigate and edit semantic nodes (like symbols, calls, modules) without having to parse entire files.

• Precise Edits & Validated Refactors: The graph model enables targeted edits with pre-defined checks, such as graph hashes and expected values, ensuring refactors are validated by the compiler.

Experimental Nature

• Breaking Changes: The language is experimental, and breaking changes are expected. Users are advised to run it in safe, isolated environments to avoid potential security vulnerabilities.

• Source-Backed Graph: ProgramGraphs are derived from source text and focus on being a tool for inspection and interchange rather than replacing primary project files.

• Version-Matched Skills: The compiler includes guides that align with the current language version to assist users in utilizing the language effectively.

Getting Started

Users are encouraged to install the ZeroLang compiler, run examples, inspect graphs, and test the edit loop to provide feedback, particularly regarding how agents can work with less uncertainty.