Here’s the **concise, command-first** workflow we agreed on (using the **rename/no-force-push** path):

```bash
# 1) Find base commit (where feature diverged from main)
BASE=$(git merge-base main feature)

# 2) Create clean branch from base
git checkout -b feature-clean "$BASE"

# 3) Cherry-pick/group commits (example: group A+B, keep C; group D+E+F, keep G)
git cherry-pick -n A
git cherry-pick -n B
git commit -m "Feat: X — initial implementation (A+B)"
git cherry-pick C
git cherry-pick -n D
git cherry-pick -n E
git cherry-pick -n F
git commit -m "Feat: X — refinements (D+E+F)"
git cherry-pick G
# (Or: pick everything then `git rebase -i "$BASE"` to squash/fixup selected ones.)

# 4) Verify equivalence
git range-diff feature...feature-clean
# (Optional sanity: git diff feature..feature-clean)

# 5) Publish clean branch & preserve old main
git push -u origin feature-clean
git push origin main:main-old-2025-08-24
```

**Then (platform UI):**

1. Set **default branch** to `feature-clean`.
2. Rename `feature-clean` → `main`.
3. (If not done in step 5) rename old `main` → `main-old-2025-08-24`.
4. Reapply branch protections/CI rules on the new `main`.

**Teammates (already cloned) quick sync:**

```bash
git fetch --all --prune
git switch main || git checkout -b main origin/main
git reset --hard origin/main
# For feature branches based on old main:
git rebase --rebase-merges --onto origin/main origin/main-old-2025-08-24 <my-branch>
```