gokeshenzhen/awesome-formal-verification-skill

This project packages up specialized knowledge about formal verification, a technique chip designers use to mathematically prove that a piece of hardware design behaves correctly, so that AI coding assistants like Claude Code, Codex, Gemini CLI, and Cursor can use that knowledge directly instead of a person having to re-explain it every time. The core idea is that formal verification, especially a specific style called JasperGold Formal Property Verification, involves a lot of specialized concepts, engine tuning tricks, and a scripting language called TCL that most AI assistants do not already know well. Rather than a person repeatedly typing out background explanations in a chat window, this repository stores that knowledge as structured, plain Markdown documents under a knowledge folder, then uses small adapter files to plug that knowledge into whichever AI agent someone happens to be using. Setting it up is a single command after cloning the repository: an install script that detects which supported AI agents are present on a machine and automatically links this repo's skill folder into each agent's own configuration location. Because the connection is a symlink rather than a copy, pulling new updates to this repository instantly updates the knowledge available to every connected AI agent, without needing to reinstall anything. The knowledge itself is organized by topic, starting with property writing, proof engine tuning, complexity management, TCL command references, and an end to end workflow guide, each one labeled with a maturity level showing whether it was extracted straight from official documentation or has actually been validated by engineers on real projects. The README lists CDC and RDC verification, a linting tool called Superlint, and support for a second EDA tool called VC Formal, as work still to come. This is a niche, specialist tool. It is aimed squarely at chip design and verification engineers who already use formal verification tools and want their AI coding assistant to understand that domain, rather than at general software developers or non-technical readers. It is released under the MIT license.