rohitg00/pro-workflow at 2.3K Stars: Self-Correcting Claude Code Memory That Compounds Over 50+ Sessions

GitHub·medium signal

rohitg00/pro-workflow (2,271 stars) packages context engineering, parallel git worktrees, agent teams, and 17 battle-tested skills around a self-correcting memory that learns from user corrections over 50+ sessions. It targets builders who want Claude Code to improve run-over-run rather than repeat mistakes. The compounding-correction framing matches the demand for self-improving, multi-session agent workflows.