correction-capture

Capture corrections as reusable coding guidelines by detecting when an engineer redirects a task, extracting a general rule from the feedback, and appending it to a shared knowledge log for future sessions.

Repeating corrective feedback happens because institutional knowledge is lost between sessions; storing these corrections as reusable lessons in a central log ensures guidelines persist and inform future work automatically.

AI learning from corrections works by intercepting task redirections, extracting the general coding rule from the specific feedback, and enforcing the loading of these learned rules in future contexts.

Store lessons learned during code reviews by appending the extracted correction and its general rule to a central tasks/lessons.md file, ensuring cross-task learning and consistency in future sessions.

No, this correction capture workflow operates without external dependencies, relying solely on a local tasks/lessons.md file to store and enforce knowledge capture rules across sessions.

Automated knowledge capture for coding guidelines is not suitable for subjective preferences or one-off context that cannot be generalized into a reusable rule for future tasks.