This repository contains a specialized environment and toolkit for AI-assisted Linux kernel development and review. While currently oriented toward Gemini, it is designed to be compatible with other AI tools and models, such as Claude. The toolkit helps developers and maintainers adhere to strict upstream Linux kernel standards by automating parts of the review and verification workflow.

• Maintainer Persona: AGENT.md defines the persona and standards for the AI agent, ensuring code and commit messages meet the high bar of the Linux kernel community.
• AI-Assisted Review: Includes ai_review.py to automate code reviews using specialized prompts.
• Subsystem-Specific Prompts: A collection of review prompts in review-prompts/ tailored for various kernel subsystems (block, mm, net, sched, etc.). Authored by Chris Mason (https://github.com/masoncl/review-prompts/tree/main).
• Semantic Code Analysis: Integrates with semcode (built via setup.sh) for semantic code search and understanding. Cloned from https://github.com/masoncl/semcode-devel .

• Linux environment
• Rust toolchain (for building semcode)
• Python 3

Run the setup script to initialize submodules and build the necessary tools:

./setup.sh

• AGENT.md: The core system prompt defining the developer persona and guidelines.
• eng.md: Detailed guidance on how to create and validate kernel changes.
• ai_review.py: Script to drive the review process.
• review-prompts/: Markdown files containing system prompts for AI-driven reviews.
• semcode/: Source code for the semantic analysis tool.
• linux/: A clone of Linux kernel source tree. Created by setup.sh.