aif-fix

Automating bug fixing requires a plan-driven workflow that applies patches and validates them against failing tests. This Skill supports immediate fixes or plan-first approaches, automatically generating patches and proposing test coverage to prevent regressions.

A plan-first approach for code patching generates a structured FIX_PLAN.md before applying changes, ensuring the bug resolution is validated. This workflow logs activities in a .ai-factory folder and proposes safeguards against regressions through test coverage.

To prevent regressions when applying automated code patches, the workflow always proposes test coverage and performs post-fix reviews. It generates patches in a designated folder and supports optional verification to validate the applied changes against existing tests.

Generating fix plans and logs requires a .ai-factory folder in your project. This directory stores the FIX_PLAN.md, execution logs, and automatically generated code patches required by the structured bug fixing workflow.

Executing an immediate bug fix is supported through a dual-mode workflow that bypasses initial planning. You can select this mode via interactive commands, allowing the system to apply and validate the code patch directly without an existing FIX_PLAN.md.

Interactive problem understanding for code bug fixes uses decision prompts to clarify the issue before patching. The workflow interactively asks questions to define the bug, ensuring the generated fix plan and subsequent patch accurately target the reported problem.