Just use symlinks and be agent agnostic (matanabudy.com)

In a recent post, an innovative approach to overcoming the fragmentation in AI coding tools was proposed, highlighting the use of symbolic links (symlinks) to create an agent-agnostic setup. With so many rapidly evolving coding agents like Codex and Gemini, developers often feel tethered to specific tools due to the lack of standardized ways to share setups and configurations. This has led to a sense of lock-in, where switching to a more advanced model means losing valuable skills and instructions tied to the original platform. The author suggests a straightforward solution by organizing all configurations and instructions in a central directory (e.g., ~/.ai/) and employing symlinks to connect these resources to various coding agents. This method allows users to maintain a common source of truth while seamlessly integrating with different tools, thus promoting flexibility and efficiency. By leveraging symlinks, developers can easily adapt to new models without being hindered by the specific demands of each tool, fostering a more cohesive and collaborative AI/ML environment.