Forge Correction Tracking
Captures and logs user corrections to continuously improve AI development workflows and templates.
The Correction Tracking skill automatically identifies instances where a user redirects, corrects, or refines Claude's output during the engineering process. By documenting these 'correction moments' in a structured retrospective format, the skill identifies recurring gaps in templates, context, or logic. This creates a powerful feedback loop that allows the Forge framework to mature over time, ensuring the AI assistant aligns more closely with project-specific requirements and team standards through data-driven improvements.
주요 기능
01Categorization of gaps in templates, context, and project assumptions
021 GitHub stars
03Automatic detection of correction signals like 'no, actually' or 'that's not right'
04Seamless integration with the forge:improve optimization workflow
05Incremental status tracking for processed and deferred improvements
06Structured logging to retrospective-notes.md with root cause analysis
사용 사례
01Refining project-specific documentation templates based on recurring missing sections
02Identifying and correcting persistent AI misunderstandings regarding architectural layers
03Building a long-term improvement backlog to mature team development standards