ls-wf-repository-localsetup-conversion

Use this workflow when onboarding a repository that may already contain old LocalSetup files, adapter paths, lockfiles, or framework source.

Every conversion starts with a report-only run. Use one shell session so the report and apply steps share the exact target and backup paths. Do not add --yes to this first command:

target_root="$(git rev-parse --show-toplevel 2>/dev/null || pwd -P)"
backup_dir="$target_root/.localsetup/backups/conversion-$(date -u +%Y%m%dT%H%M%SZ)"
localsetup convert --target-directory "$target_root" --backup-dir "$backup_dir" --tools codex --packs core

Record the report output and stop unless it provides all of this evidence:

• blockers are empty; unmanaged or ambiguous content is a blocker
• the reported LocalSetup source checkout and target repository root are the intended roots
• the report names the same timestamped backup_dir and its conversion-report.json path

Only after that evidence is explicit may you run the applying command with the unchanged variables:

localsetup convert --target-directory "$target_root" --backup-dir "$backup_dir" --tools codex --packs core --yes

After apply, confirm the recorded backup and report exist, then run the verification and doctor commands in ls/docs/REPO_CONVERSION.md. Follow that guide for source-vs-target behavior, blocker handling, and rollback evidence.