aif-evolve

To make AI adapt to project-specific conventions, you need a self-improvement mechanism that analyzes past patches and codebase patterns to update AI rules. This skill automatically identifies recurring errors and adds mandatory guards to existing AI skills.

AI skill evolution is the process of enhancing AI capabilities by analyzing accumulated patches and codebase patterns. It improves codebase analysis by identifying recurring issues, extracting project-specific conventions, and systematically updating skill definitions with new rules and guards.

You can automatically update AI skills by running a skill that analyzes your project context, accumulated patches in `.ai-factory/patches/`, and existing definitions in `.github/skills/`. It identifies recurring mistakes and rectifies them by adding project-specific rules to skills like `/aif-fix`.

Yes, you need existing AI Factory skills and a description file to use aif-evolve for pattern recognition. The skill requires analysis of `.ai-factory/DESCRIPTION.md`, patches in `.ai-factory/patches/`, and existing skill definitions in `.github/skills/` or `skills/` to function.

The best way to prevent recurring null check errors is to analyze past patches to identify the pattern, then automatically update your AI code generation skills to include a mandatory null-check guard, preventing future similar bugs across the project.

Your AI assistant ignores project-specific architecture patterns because it lacks contextualized rules. By analyzing your codebase patterns and accumulated patches, a self-improvement skill can adapt AI guidance to your tech stack and inject specific guards into skill definitions.