Show HN: Aver – a Language Designed for AI to Write and Humans to Review

Aver introduces a new language designed with a specific purpose in mind: making code generated by local LLMs more readable, reviewable, and auditable by humans. Rather than asking models to generate complex Python or JavaScript, Aver provides a syntax optimized for AI comprehension and human verification, reducing friction in workflows where AI coding assistants augment human developers.

For organizations deploying local LLMs for code generation—whether through Ollama, vLLM, or other self-hosted inference engines—Aver offers a middle ground between raw natural language and fully executed code. The language design prioritizes clarity and verifiability, making it easier for human reviewers to spot errors, understand intent, and approve or modify generated solutions before execution. This is particularly valuable in regulated environments or teams with strict code review requirements.

Aver is available on GitHub and represents an interesting approach to human-AI collaboration in code generation, where the tooling itself is optimized for the unique characteristics of machine-generated output rather than forcing models to conform to languages designed for humans.

Source: Hacker News · Relevance: 6/10