aif-fix

What problem does it solve? Fixing bugs with an AI assistant often means vague patches, no reproduction of the original failure, and no record of what was learned. This Skill enforces a disciplined bug-fix workflow: reproduce the problem with a minimal regression check, implement the smallest root-cause fix, verify the same check passes, and capture the lesson in a reusable patch file. ## Core Features & Use Cases - Regression-First Fixing: Identifies or creates the smallest regression check, confirms it fails before the fix, and reruns it after to prove the fix works. - Two Modes: Choose "Fix now" for immediate remediation or "Plan first" to generate a reviewable FIX_PLAN.md before any code changes; running with no arguments executes an existing plan. - Learning Patches: Every fix produces a timestamped patch file documenting the problem, root cause, solution, and prevention guidance, building a project knowledge base over time. - Use Case: You hit "TypeError: Cannot read property 'name' of undefined in UserProfile". Invoke the Skill with the error message; it locates the code, writes a failing regression test for the null case, applies a logged fix, verifies the test passes, and saves a patch so the same class of bug is prevented later. ## Quick Start Ask the assistant to fix a bug by running the aif-fix skill with your error message, for example: fix the bug where the orders API returns an empty array for authenticated users.