process-feedbacklisted

# /process-feedback Process a feedback file the user has written. Two passes are mandatory - both must complete. **Usage:** `/process-feedback <filepath>` --- ## Step 0 - Rename the feedback file Before reading the file, rename it to include the short commit SHA of the repo's latest commit. This ties the feedback to the exact codebase state it reflects. ```bash cd <repo-dir> git log --oneline -1 HEAD # gives e.g. "bd4c939 fix: ..." # Rename file from feedback.txt -> feedback_bd4c939.txt (use full 7-char short SHA) ``` Skip this step if the file is already named with a SHA (e.g. `feedback_bd4c939.txt`). After renaming, **commit the renamed file immediately** before reading or processing it: ```bash git add docs/Development/Feedback_*.txt # or whatever path git commit -m "chore: rename feedback file with commit SHA" ``` This preserves the raw feedback in git history before it is deleted. --- ## Pass 1 - Product tasks Read the file. Extract all product bugs, feature requests, and UX issues. Route each item: - Bugs / features for an active project -> that repo's `docs/IDEAS.md` (quick wins first) - Cross-repo tasks -> the cross-repo pending-actions file - Completed fixes already in the codebase -> skip or note in HISTORY.md **Optimization note:** Use ONE bash command to locate IDEAS.md: `cd <repo-dir> && find . -name "IDEAS.md" -type f`. Read the result directly. ## Pass 2 - Claude learnings (MANDATORY - skill fails if skipped) Re-read the same file with a diff